Author Topic: Relevance of polarized orbital  (Read 2564 times)

0 Members and 1 Guest are viewing this topic.

Offline CVD

  • New QuantumATK user
  • *
  • Posts: 3
  • Reputation: 0
    • View Profile
Relevance of polarized orbital
« on: September 21, 2009, 16:02 »
Hi,

I've got a little question about polarized orbitals. I think using this kind of orbitals in the basis set can be relevant or not in accordance with the element and the configuration. So, I'm wondering for a criteria to decide if we have to use polarized or not for the elements in my configuration.

My idea for now is to look at the output file with high verbosity after the convergence of a polarized self-consistent calculation :
104  C     q =  3.86388 [ s:  0.562, s:  0.478, py:  0.631, pz:  0.337, px:  0.525, py:  0.302, pz:  0.507, px:  0.401, d(-2):  0.012, d(-1):  0.028, d( 0):  0.025, d( 1):  0.032, d( 2):  0.022 ]
(results from a doublezetapolarized calculation)

We can see by making the sum of the square of each coefficient that the polarized part is only 0.166% of the total. So in this case I want to say that using polarized orbitals for this atom is unnecessary.

Do you think that my reasoning is right ? What's the minimal percentage to say that's polarized is unnecessary ?

Offline Nordland

  • QuantumATK Staff
  • Supreme QuantumATK Wizard
  • *****
  • Posts: 812
  • Reputation: 18
    • View Profile
Re: Relevance of polarized orbital
« Reply #1 on: September 21, 2009, 16:51 »
The polarization orbital is almost always an unoccupied state, but the idea is in principal good.
However the presence of a polarization orbital changes the Hamiltonian -despite it being unoccupied, and therefore it will affect your results.

The best way to investigate if the polarization orbital is needed, is to perform a calculation of a property and see if it depends on polarization orbital or not.
For instance if you want to find out, if it is needed to use the polarization orbital for Gold, perform a equation of state calculation, calculating the total energy for a series of lattices constants. If the energy curve shows a significant different structure, the you should go with the polarization orbital.
If your are interested in transport, I would suggest performing a calculation on a gold wire to see the effects of the polarization orbital in the transmission spectrum.

For some system it is needed strongly to use the polarization orbital to get any good results, and for other systems it is simply a waste of calculation time.