3

When considering $\mathrm{DFT}+U$ calculations, people either go with (1) first-principles approach: calculating the $U$ parameter using linear response theory, $\mathrm{DFPT}$, $\mathrm{ACBN0}$, etc., or (2) empirical approach: tuning the U parameter to reproduce experimentally known band gaps.

I am interested if there is any experimental method which can directly, or indirectly estimate the Hubbard $U$ parameter that is effective in the material. We do not then need to run calculation varying $U$ values and neither we need to calculate $U$ from expensive first-principles approaches. We can simply analyze the experimentally available data to have an estimate of the $U$ values and use that in our $\mathrm{DFT}+U$ calculation.

AbdulMuhaymin
  • 223

2 Answers2

1

$U$ is a model parameter. To estimate it all you can do is to predict measured properties of your physical system and then look for the $U$ value that best explains them.

my2cts
  • 27,443
1

When considering $\mathrm{DFT}+U$ calculations, people either go with (1) first-principles approach: calculating the $U$ parameter using linear response theory, $\mathrm{DFPT}$, $\mathrm{ACBN0}$, etc., or (2) empirical approach: tuning the U parameter to reproduce experimentally known band gaps.

I am interested if there is any experimental method which can directly, or indirectly estimate the Hubbard $U$ parameter that is effective in the material.

The Hubbard $U$ is a toy model parameter. You can not really measure it "directly."

There are experimental techniques like ARPES, RIXS, and x-ray absorption that can be sensitive to electron-electron interactions and thereby can be related to model parameters like $U/t$. For example, here is an ARPES paper that purports to provide values of $U/t$.

We do not then need to run calculation varying $U$ values and neither we need to calculate $U$ from expensive first-principles approaches. We can simply analyze the experimentally available data to have an estimate of the $U$ values and use that in our $\mathrm{DFT}+U$ calculation.

I don't know if it is going to be as "simple" as you think, but sure, go for it.

Remember too that DFT calculations are just self-consistent single-particle calculations regardless of how fancy you get with your potential... So it is quite difficult, if not impossible, to capture the effects of electron-electron interaction in such a theory. In other words, a self-consistent piece of garbage is still a piece of garbage.

hft
  • 27,235