Re: How to determine FOLP parameters?

[Hiroshi Oji <oji-a@st.ritsumei.ac.jp>]

Thank you for your reply, and I apologize for my rate response.

Michel Jaouen wrote:
> > Dear FEFF users,
> >
> > I am trying to calculate F K-edge XANES of some hydrogen containing solid
> > (several model compounds imitating partially fluorided hydroxyapatite) by
> > FEFF8.20.
> >
> > Some of the calculation was done without trouble.  However in a 
> > certain model (
> > see the attached file "feff_inp_1.txt"), the calculation stops with 
> > following
> > error messages:
> >
> >   WARNING: fatal error in subroutine corval. Try
> >     to reduce ca1 in SCF card. If does not help,
> >   SEND bug report to AUTHORS
> >   CORVAL-1
> >
> > (Please see also the attached file "log1_dat_1.txt".)
> >
> > Following the advice in the error message, I changed ca1 in SCF card 
> > to 0.1 or
> > 0.05, but unfortunately it did not work.
> >
> > Then I changed the folp parameter in FOLP card (determining the 
> > magnitude of
> > overlapping the muffin-tin radii) as follows,
> >
> >   FOLP  4  0.8
> >
> >  and actually it did work.  (If I am correct, this option means folp 
> > for the
> > other potentials are all set to be 1.0. )
>
>
> I don't think so : for the other atomic species it remains to the 
> default value (1.15).

I tried following three options,

(1)
 FOLP 0 1.0
 FOLP 1 1.0
 FOLP 2 1.0
 FOLP 3 1.0
 FOLP 4 0.8

(2)
 FOLP 0 1.15
 FOLP 1 1.15
 FOLP 2 1.15
 FOLP 3 1.15
 FOLP 4 0.8

(3)
 FOLP 4 0.8

The results of (1) and (3) were exactly the same, but different from that of 
(2). According to xmu.dat, the Muffin tin radii (Rmt in xmu.dat) for (1) and (3) 
are set to be equal.  So I still think I'm correct.

> > However, I doubt this results because the shape of simulated XANES for 
> > the other
> > model by default folp (AFOLP) is significantly different from that by 
> > modified
> > folp (FOLP 4 0.8).
> >
> > I am wondering if there are "strategies" for determining appropriately 
> > the folp
> > parameters.  If there is, could someone kindly give me the advice?
>
>
> No general strategie, look at the appendix of the paper of JJ Rehr and 
> RC Albers (Reviews of Modern Physics, Vol72, n°3, July 2000, p. 621) who 
> discussed about that. But be happy with your result: in fact Pot 4 is 
> for H if I read your input file, and H is a small atom: so why not folp 
> 0.8 for H? But look at the charge transfer to check is there is some 
> sense from the physical point of view.

Thank you for telling me the reference.  According to this paper, it seems to be 
there is no straight-forward way to determine the radii of muffin-tin.

Following your advice, I checked the charge transfer and found that the charge 
transfer for Ca is around 0 - 0.4 for all the models.  I believe Ca is almost 
completely ionized in apatites, i.e., the charge transfer for Ca should be about 
+2.0, so the folp parameter I used might not be good.  Am I correct?

Best regards,

Hiroshi Oji


> Best Regards,
>
> M. Jaouen

--
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
  Hiroshi Oji, Dr.
  SR Center
  Ritsumeikan University
  1-1-1  Noji-higashi
  Kusatsu
  Shiga 525-8577
  JAPAN
  E-mail oji-a@st.ritsumei.ac.jp
<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<