NEUTRAL & CHARGE DEFECTS

Message

giacomo giorgi · #1 Post by **giacomo giorgi** » Fri Mar 13, 2009 10:04 am

Dear all,
I am trying to get some feedback with defects in bulk. Honestly I do not know if I am moving towards the right direction. I read the manual (Dipole corrections for defects in solids) but I still have trouble.
Here are my doubts.

I have a 64 atom cell of GaAs and I want to replace one Ga with 1 Si, studying it first as a neutral, then as charged -1.

As neutral case,
1) I optimize my neutral cell (with Si replacing Ga) and I get my energy (in principle there should be not so large effects due to dipole corrections since neutral system).

As a charged defect
2) I want to study the system charged -1. Then I input a new NELECT (= initial + 1e-). Then I optimize. (the manual states to calculate first energies WITHOUT corrections)

3) To correct the dipole/quadrupole, and energy, I restart setting DIPOL= (coor of the def) and LDIPOL=.TRUE.

Since I get two TOTEN values, one from 2) and one from 3), my question is:

Must the value I get for the corrections (dip/quad/Energy) be added by hands (as I read in the manual) on the energy calculated in 2)

Or the TOTEN I have to take into account is DIRECTLY that calculated 3)

I am honestly in trouble.... The more I read messages in the forum and the manual the less I am able to give me an answer.

Please help me,
thanks!
G

#2 Post by **admin** » Wed Mar 18, 2009 12:54 pm

for the higher multipole corrections, please keep in mind that
-- you have to check first (by having a look at the charge density difference of the perfect and the defect cell) whether the charge is localized (in the defect)
-- as written in the manual, just set IDIPOL=4; DIPOL (coordinates), and LDIPOL=.False. because the automatic correction is not possible in vasp.4.6,
-- then calculate the defect induced quadrupoles by taking the difference of Tr[quadrupole] of the perturbed and the unperturbed cell. rescale the contributions to the multipole corrections according to the manual
(dipole-quadrupole, monopole-quadrupole), add the monopole-monopole contributions and finally re-scale with the multipole corrections with dielectric constant ε. It can be found in standard reference handbooks for many solids. vasp.4.6 does not offer the calculation of ε from ab initio.
concerning TOTEN as written in OUTCAR: the very last TOTEN written for each ionic step (i.e.after electronic convergence is achieved) includes the dipol energy. You can easily check this contribution by "grep TOTEN OUTCAR"

hra · #3 Post by **hra** » Fri Apr 10, 2009 5:03 am

Bear in mind that once you change NELECT (or put a defect in your system), your eigenvalues are shifted, which requires further corrections. I prefer to figure out the potential alignment and image charge corrections by hand, and forget about built in tags like IDIPOL et al.

giacomo giorgi · #4 Post by **giacomo giorgi** » Mon Apr 13, 2009 3:51 am

Thanks a lot. But could you kindly be more explicit? Can you explain me how to do it? Are there any restrictions for this procedure?
I am treating 1 Ge substitituting Ga and As at once, respectively, and on this system I am varying the charge (NELECT) from -2 to +2. How to correct by hands charge effect? Must the defect still localized? My very best and thanks in advance,
G

hra · #5 Post by **hra** » Tue Apr 14, 2009 3:35 pm

An explicit and exhaustive explanation is beyond the scope of this forum

See e.g. PRB 78, 235104 (2008).