Re: Feff8 questions

["John J. Rehr" <jjr@leonardo.phys.washington.edu>]

Dear Ray,

 Thanks for your email and questions on how to use FEFF8.

> 1. We are aware that your theoretical method gives broadened DOS. Also we 
> noted that it is possible to take into account the experimental resolution by 
> using the CORRECTIONS card. Our question is how to properly incorporate the 
> experimental resolution?. Should we put directly the value of our resolution 
> in the card or do we have to consider the difference between our resolution 
> and the broadening inherent in the method? (is there some way to estimate this 
> intrinsic broadening?).

  The intrinsic lifetime broadening Gam_ch is taken into account by
default in FEFF8 - it's taken from tables (see the FEFF references for
details) and printed out in eV in the headers of say xmu.dat:

  You can add additional experimental broadening as V_i in CORRECTIONS or
EXCHANGE cards. Please see the FEFF8 doc for details/examples.


%head xmu.dat
# Al Metal cfc a=4.049 from G. Hug                               Feff 8.20
# POT  SCF  30  4.1000   0, NO core-hole, AFOLP (folp(0)= 1.150)
# Abs   Z=13 Rmt= 1.531 Rnm= 1.573 L3 shell
# Pot 1 Z=13 Rmt= 1.531 Rnm= 1.573
# Gam_ch=3.628E-02 H-L exch
# Mu=-7.984E+00 kf=1.825E+00 Vint=-1.667E+01 Rs_int= 1.988
# FMS rfms= 7.6000
# PATH  Rmax=-1.000,  Keep_limit=  .00, Heap_limit  .00  Pwcrit= 2.50%
#  S02=1.000                                        Global_sig2=  .00000
#  Curved wave amplitude ratio filter   4.000%
#     file         sig2 tot  cw amp ratio   deg  nlegs   reff  inp sig2
#     0/   0 paths used
#  xsedge+ 50, used to normalize mu           3.2095E-02
#  -----------------------------------------------------------------------
#  omega    e    k    mu    mu0     chi     @#

> 
> 2. We are simulating EELS spectra and we are interested in make 
> orientation-dependent simulations with FEFF. For very small scattering angle, 
> the momentum transfer from the electron is in the incident direction and we 
> need your advice about how to set the polarization in the POLARIZATION card 
> and the ELLIPTICITY card. For example, if the electron beam is going to the 
> crystal along the z axis (001 direction), what should be the indexes in both 
> cards? (probably 001 in both?).
   the ellipticity  is along k so this vector is 001
   polarization vector epsilon can be any vector perpindicular to k
     choose epsilon to correspond to your experimental setup.

if neither polariation or ellipticity is specified FEFF does a polarization
average.

Please see the feff8 doc for details examples.

> 3. The final point is about how to do the scaling of different edges. We are 
> simulating tin oxides. The edges involved are the M4 and M5 of tin and K-edge 
> of oxygen. We understand that feff gives the absorption per electron, is that 
> right?. Then, the immediate question that arises is: should we take in 
   No - I don't think so.
> consideration the degeneracy of the excited level for the proper scaling of 
> the edges?. Also, for a compound should we include a factor which accounts for 
> the stoichiometry ?

  FEFF by default gives a normalized absorption  mu/mu_0_50
where the constant mu_0_50 is the absorption 50 eV above the edge.
To get the absolute absorption in Megabarns, multiply the mu data
in xmu.dat (col 4) by the normalization constant mu_0_50 given
in the last line of the header, e.g. 3.2095 E-02.

   Thus if you want to add M4 and M5 edges, you need to multiply
the M4 data by the constant for the M4 edge and the M5 data by
the corresponding M5 value.