Unable to do the spin transport Calculations in organic inorganic perovskites

[gayani2025]

Dear All,

I have been actively exploring spin transport simulations in QuantumATK, referring to several tutorials and incorporating valuable insights provided by the ATK community. Attached are two snapshots of my current system setup, where I aim to investigate spin transport properties through a perovskite device region, using Ni and Au as electrodes.

I have fully optimized the bulk structures of Ni, Au, and the perovskite, and constructed the interface regions using the Interfaces tool in the Builder. However, based on my current understanding and experience with this system, it appears that performing spin transport calculations in this configuration may not be feasible.

As evident in the attached snapshots, the system suffers from relatively high interfacial strain and involves a large number of atoms in the device region. Additionally, the nature of the heterojunction requires longer electrodes for accurate transport results, which would further increase the system size significantly and add to the computational complexity.

To the best of my knowledge, I have not come across any published spin transport studies involving perovskites in this specific configuration, which suggests that this may be a particularly challenging system to model.

I would be very grateful for any suggestions or feedback on how best to approach this problem, including potential alternatives or simplifications that could make the study more computationally tractable.

Thank you for your time and consideration.

GN

[Anders Blom]

A few hundred atoms, even 1000, really isn't that much if you have a good enough resource to run on, although yes spin will add complexity. I would go for a smaller basis set like FHI/SingleZetaPolarized, which should be accurate enough to capture the broad features of this system.

[gayani2025]

Thank you very much for your valuable comment. I also wanted to ask whether you have come across any example studies on spin transport in perovskite systems of this type. Despite searching the literature, I was unable to find relevant examples. If you happen to know of a study involving spin transport in perovskites—particularly with similar device configurations—it would be greatly appreciated if you could share it.

Additionally, I have access to a computing cluster where I can use up to 64 CPUs. I would be grateful if you could comment on whether this computational capacity is likely sufficient for a spin transport simulation of this scale, or if I should expect significant limitations.

Thank you again for your time and guidance.

[Anders Blom]

Memory is the limiting factor, not the number of CPUs. I am not directly familiar with studies like this but I don't see any principle reasons it would not work.

[gayani2025]

Hi,

Thank you for your message. To overcome the computational limitations, we have designed a simplified system using only the chiral molecule for the transport calculation, aiming to capture the qualitative spin selectivity.

I have set up the geometry optimization of the central region following the available tutorials. I believe this setup should capture the essential features of the system.

I would appreciate it if you could have a quick look and let me know if the current setup looks reasonable.

Thank you again for your support.