Request for Feedback on Spin Transport System Setup

[gayani2025]

Dear All,

I am currently setting up a system to study spin transport using an organic-inorganic perovskite structure. The electrodes I am using are Ni (for spin injection) and Au.

After reviewing some relevant tutorials and incorporating suggestions from earlier discussions, I have outlined the following proposed steps for the study:

Optimize the device configuration

Extend the electrode lengths as needed

Perform the spin transport calculation

I have already prepared the geometry optimization script for the initial step. As a preliminary test (Test 1), I have assigned spin polarization as follows: Ni with spin-up, and perovskite and Au with zero spin, based on discussions with our experimental collaborators.

I would appreciate it if someone could kindly review my parameter selections and let me know if they are appropriate for this system.

Thank you very much for your support.

Best regards,
GN

[AsifShah]

Hi,

Just my few cents:
If you are optimizing this configuration as bulk, you will need to add a vacuum in the z direction (to avoid spurious interaction between Ni and Au) and keep either electrodes (say Ni) outer layer fixed to emulate bulk behavior. At the same time, you need to keep the gold outer layer rigid (or fixed in xy direction) so that it can move in the z direction to avoid any internal stresses during optimization. This way, you can optimize the structure.
Also, make sure the gold and nickel are optimized from the bulk unit cell before forming the interface. Also, choose a low-strain interface from Interface Builder (though it depends on computational complexity).

Also, I wanna know if initial spin has any effect on the final optimized configuration or just helps in quick optimization?

[gayani2025]

Hi,

Thank you very much for your comments.
I will first do the optimization of the electrodes before making the system using interface builder. I have a concern on adding vacuum on z direction. I made this system following https://docs.quantumatk.com/tutorials/fe_mgo_fe/fe_mgo_fe.html tutorial. However they have not included a vacuum slab. Please explain this to me.

Thanks
GN

[AsifShah]

Hi,

The tutorial you referred to uses a device configuration where you don't need to add a vacuum in the z direction; it takes care of it internally. However, the configuration you have shown in the pic is a bulk configuration. To optimize it as is given in the tutorial, you need to first convert your bulk configuration into a device, which you can do in the builder. In the builder, select your configuration from the stack, and then on the right, choose Device from the bulk.

[gayani2025]

hi, Thank you very much and now I understand what you meant. I think I misunderstood the procedure. Now I'm in the process of optimizing the  Ni (111) and Au (111) separately. I will optimize the perovskite as well before making the electrode structures. In this optimization procedure i have not select the cell length optimization. Please see the attached script. Please correct me if that is wrong.
Thanks
GN

[AsifShah]

Hi,
You don't need to use a vacuum when optimizing the bulk Ni or gold unit cell. Just optimize the bulk unit cells and then cleave 111 out of it.

Also, I see you have chosen a very less K-point sampling. The default for Ni is 12x12x1, and you have used 6x6x1.
Also, use a density mesh cutoff of around 110 Hartree.

The actual way to figure out these numbers is to calculate the total energy of the unit cell at various density mesh cutoff and K point sampling, and see at what values the total energy converges. Generally, the larger the supercell, the less K point sampling is required, and the smaller the cell, the larger the K point sampling.
Also, whether you want to optimize with/without cell optimization depends on which one gives you the right lattice parameters. This can you verify with literature or experiments.

[gayani2025]

Thank you for the detailed explanation. So I guess here the perovskite also I should start with the Bulk optimization right?