[Ifeffit] Problems with potential overlap for small molecule

[Renske van der Veen]

Dear FEFF users,

I have a FEFF8.4-related question concerning the abortion of the potential
module in a XANES calculation for a small molecule in solution. It might not
be totally appropriate for the ifeffit mailing list, but the FEFFusers
mailing list doesn¹t seem to be used very often. I have looked through the
archive, but I haven¹t found an entry with this specific problem.

I am trying to simulate the L3-edge XANES spectrum for the linear I3-
molecule (I-I-I) embedded in a solvent with O-H groups. In order to simulate
the scattering from the solvent, I have put two O-H groups at either end of
the molecule with the H-atoms pointing towards the I-atoms (expected charge
transfer from I3- towards solvent). The I-H and I-O distances are taken from
ab initio calculations in the literature, the O-H distance is well-known and
amounts to 0.96 A. It is also known from the literature on similar ions in
solution that the H-atoms contribute to the XANES and should therefore be
included.

I pasted my feff.inp file below.

Running this file results in abortion at the beginning of the potential
module:

...
Number of processors =            12
 Calculating potentials ...
    free atom potential and density for atom type    0
 MOVRLP-1
 Abnormal termination on processor             8
[n7:09539] MPI_ABORT invoked on rank 8 in communicator MPI_COMM_WORLD with
errorcode 512
...

It seems to occur when it is trying to overlap the potentials (MOVRLP-1). I
tried including the FOLP 2 0.8 card as suggested for H-atoms in the manual,
but this does not help.

It seems like the O-H bond <1 A is too small: moving the O atoms further
away such that the O-H bond is 1.05 A ³solves² the problem, although this is
not what I want. But: moving the H atom closer to the I atom while keeping
the O positions fixed, which increases the O-H bond as well, does not help!
Also, to make things even more complicated, excluding the H-atoms fully,
i.e. putting only two O-atoms at 3.5 A from the iodines, results in the
WARNING during the SCF loop that the interstitial potential level lies
higher than the Fermi energy (moving the O atom closer to ca. 2 A removes
this message). In addition, certain combinations of H- and O- positions give
the output ³negative dens² during the SCF loop, for which I checked that
FOLP 2 0.8 did not help...

It is obvious that FEFF has problems with generating the MT potentials for a
small and ³open² system such as the one here. Did somebody encounter similar
problems (abortion, Ei>Ef, negative dens) for systems like this? Is there
something wrong with my input cards?

I¹d appreciate any help.

Best regards,
Renske van der Veen

P.s. I am running the parallel version of FEFF8.4 on Linux

***

TITLE  I3- XANES spectrum

*  I L3 edge energy = 4557.0 eV
EDGE L3
S02 1.0

CONTROL 1 1 1 1 1 1
SIG2 0.01

PRINT 1 0 0 0 0 0
EXCHANGE 0 0.0 0.0 0
 
SCF 10 1 30 0.2 1 

LDOS -30 60 -1 

XANES 6

FMS 10 1     
UNFR

POTENTIALS
0 53 I 3 4
1 53 I 3 4
2 1 H
3 8 O 2 3

*FOLP 2 0.8

ATOMS
0.0000    0.0000    0.0000    0
-2.91    0.0000    0.0000    1
2.91    0.0000    0.0000    1
-5.45    0.0000  0.0000  2
5.45    0.0000    0.0000    2
-6.41    0.0000  0.0000  3
6.41    0.0000    0.0000    3
END

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://millenia.cars.aps.anl.gov/pipermail/ifeffit/attachments/20090520/dde0ac9f/attachment.html>