What is the high_frequency_dielectric

Does the high_frequency_dielectric is the “MACROSCOPIC STATIC DIELECTRIC TENSOR IONIC CONTRIBUTION” in the vasp OUTCAR?

Hi Zhibin,

In general, people discuss two dielectric constants:

  • The high-frequency dielectric constant (\epsilon_\infty), sometimes called the “optical dielectric constant”, is the value deduced from the refraction of electromagnetic waves with frequencies high compared to lattice vibrations (phonons).
  • The static-dielectric constant (\epsilon_\mathrm{S}) is the total dielectric screening from the high-frequency response and the vibrational response (lattice). I.e, static = high-frequency + ionic dielectric constants.

Unfortunately, the terminology used by VASP for the dielectric constants is not very precise. If you do a DFPT+LEPSILON calculation, as described in the calculation section of the documentation, the OUTCAR will contain a number of dielectric constants, each with very similar names.

For example, these are taken from the OUTCAR for a DFPT calculation on SnS.

 MACROSCOPIC STATIC DIELECTRIC TENSOR (including local field effects in DFT)
 ------------------------------------------------------
          17.540787     0.000000     0.000000
           0.000000    21.425820     0.000000
           0.000000     0.000000    17.487301
 ------------------------------------------------------

...

 MACROSCOPIC STATIC DIELECTRIC TENSOR IONIC CONTRIBUTION
 ------------------------------------------------------
          19.848112    -0.000007    -0.000001
          -0.000007    41.125796     0.000008
          -0.000001     0.000008    21.693346
 ------------------------------------------------------

Here, the first dielectric constant is actually the high-frequency dielectric constant NOT the static dielectric. The second tensor is the ionic contribution to the static dielectric. To get the final static dielectric constant you have to take the high-frequency dielectric + ionic contribution.

Therefore, the inputs to amset would be:

static_dielectric: 
  - [37.39, 0, 0]
  - [0, 62.56, 0]
  - [0, 0, 39.18]

high_frequency_dielectric:
  - [17.54, 0, 0]
  - [0, 21.43, 0]
  - [0, 0, 17.49]

I hope that helps.

Best,
Alex

1 Like

Dear Alex,

You shocked me!
I have wrongly regarded “MACROSCOPIC STATIC DIELECTRIC TENSOR (including local field effects in DFT)” as the static-dielectric constant for a very long time! I always stupid to make a subtraction from:
high-frequency = static - ionic == MACROSCOPIC STATIC DIELECTRIC TENSOR (including local field effects in DFT) - MACROSCOPIC STATIC DIELECTRIC TENSOR IONIC CONTRIBUTION.

You discovered and corrected a blind spot of my knowledge!
Thank you very much!

Best wishes,
Zhibin

A post was split to a new topic: How is the dielectric constant calculated

My understanding is that the first dielectric matrix is the electronic part and the second matrix the ionic part. At zero frequency, the static dielectric matrix of a compound is the sum of the two matrices. When the frequency approaches infinity, because ions can’t move as fast, the ionic contribution becomes zero. Therefore, at high frequency, the total dielectric matrix is equal to the electronic part.

And I think using the electronic part of the dielectric function at the zero frequency as the electronic part of the dielectric function at infinite frequency is an approximation. It probably is better to calculate the electronic part of the dielectric function as a function of frequency using the linear response theory and use that as the dielectric function at infinity.