LEPSILON for Born effective charges in metallic system

[shan_feng]

Dear VASP Developers and fellow users,

I am currently encountering some questions for which I would appreciate your clarification. As stated on the VASP official website (https://www.vasp.at/wiki/Category:Linear_response), “LEPSILON can be used for local or semi-local exchange-correlation functionals and applies to both semiconductors and metals.”

I would like to ask whether LEPSILON is applicable to metallic systems for the purpose of obtaining Born effective charges. In other words, is the calculation of Born effective charges using LEPSILON theoretically and technically meaningful for metals in VASP? Is this approach based on the adiabatic approximation? Are there any references that support the physical meaningfulness of the calculated Born effective charges in metallic systems?

Thank you very much for your time and assistance. I look forward to your response.

[alexey.tal]

Dear shan_feng,

Indeed, the DFPT implementation in VASP can be used for both semiconductors and metals.

Thus, in principle, LEPSILON can be used to calculate Born effective charges (BECs) for metals. However, the BECs are evaluated only within the adiabatic approximation, which implies that they vanish for metallic systems. Since you raised this question, you are probably aware that non-adiabatic BECs can be defined and computed for metals, and that they do not vanish, as demonstrated in Ref. here.

At present, VASP does not support the calculation of non-adiabatic BECs. We will add this feature to our to-do list.

Best regards,
Alexey

[yujia_teng]

Dear Alexey,
I have the same question here. So you said that BEC should vanish for metallic system. But I have a metallic heterostructure (metal/insulator/metal), and the BEC of each atom is non-zero. The insulator actually becomes metal here due to charge transfer.

So here, is the BEC here still a reliable and meaningful value, and can be used to compare with the BEC of the freestanding insulator?

Best,
Yujia