[Ifeffit] problem of overestimated E0 (Edge-energy)

Fri Jan 13 14:26:33 CST 2012

Hello Jeremy
When using SCF, I got smaller E0(6 eV) than before(10eV). 
Eventhough It is better, that energy is still larger than white line.  
In my way, the muffin-tin potential has been enough for EXAFS with single scattering.

I wonder which is the better between SCF and muffin-tin pottential in EXAFS, not XANES 
and why do the energy shifts to above white line in Pt-foil, PtO2 and others make?  is that  problem in FEFF calculation or theory?

Date: Fri, 13 Jan 2012 10:05:14 -0600
From: kropf at anl.gov
To: ifeffit at millenia.cars.aps.anl.gov
Subject: Re: [Ifeffit] problem of overestimated E0 (Edge-energy)

Jeong-Eun (I hope that is right),

Looking at the feff.inp file, the SCF line has been commented out.  This indicates that the self-consistent calculations have not been done.

Try:
SCF       5   

You can adjust the n_scf and ca later if this doesn’t converge.  This could shift Eo by several eV.

Jeremy

From: ifeffit-bounces at millenia.cars.aps.anl.gov [mailto:ifeffit-bounces at millenia.cars.aps.anl.gov] On Behalf Of JeongEunSuk
Sent: Thursday, January 12, 2012 11:43 PM
To: ifeffit
Subject: Re: [Ifeffit] problem of overestimated E0 (Edge-energy)

Thank Jeremy
the feff.inp file and Fourier transformed EXAFS is attached.
I fitted the nearest neighbor scattering path, only single scattering Pt
and I am sorry, 
  I wrote wrong fitting range
Fitting range
 K-sapce: 3.0 ~ 15 A-1
 R -space: 1.3~3.3  A 

Date: Thu, 12 Jan 2012 16:05:25 -0600
From: kropf at anl.gov
To: ifeffit at millenia.cars.aps.anl.gov
Subject: Re: [Ifeffit] problem of overestimated E0 (Edge-energy)

I think we’ll need more information to help out: at a minimum, the feff.inp file.  Also, are you fitting just the nearest neighbor scattering path?

To begin to answer your question, while it is often the case that Eo falls below the white line peak, this is not categorically true.  A relevant example might be PtO2.

Jeremy
Chemical Sciences and Engineering Division 
Argonne National Laboratory 
Argonne, IL 60439 
Ph: 630.252.9398 
Fx: 630.252.9917 
Email: kropf at anl.gov 

From: ifeffit-bounces at millenia.cars.aps.anl.gov [mailto:ifeffit-bounces at millenia.cars.aps.anl.gov] On Behalf Of JeongEunSuk
Sent: Thursday, January 12, 2012 3:47 PM
To: ifeffit
Subject: [Ifeffit] problem of overestimated E0 (Edge-energy)

Hello all
I want to understand EO(edge-energy), precisely. (In Artemis, EO is writed as Enor)
Generally, EO is energy when electrons start to go out from core atom in XAFS. 
As you know, the experiments show that to find EO is not easy in XAFS. 
To remove background, sometimes I chose EO, the value of the maximum derivative in Athena. 
and when fitting, the E0 is correted by theoritical FEFF. 
This is my question.
The attached file is the figure of Pt foil(Pt L3 edge measurement). 
I think edge-enegy has to be in range between A and C because electrons start to move to continum state in that range.
However, fitted E0 is shifted to C (about 11573eV). and then when removing background I chose E0=11563eV(This is Pt L3 edge).
Fitted E0 is overestimated value, compared with whit line(B).
That means the calculated EO obtained from FEFF8.0 (using auto-sel! f consistency field potential) is larg! er than that of experiment. 
 Do I need to change SCF potential in FEFF8.0?

----------------------------------------------------------------------------------------------------------------------------

This is fitting parameters and results(in my case, I used Linux version)

To remove background
    k -range = 2.5 ~ 15,   EO=11563eV, Rbkg=1.3, kweight =1

To fit data

k -range = 3.0~13.5,  R -range =2.3~13.5

1. fitting parameters

  set     macc = 0.0
  set     so2 = 0.89    // reduction factor
  guess   ePt = 0.0   // energy shift
  set   npt1=1.0                // the number of Pt (1.0  means 12&nb! sp; Pt atoms around core Pt)
 gu! ess    dpt1 = 0.0042     // effective bonding length
 guess    sigpt1=0.005100   //Debye waller factor

2. fitting results
  fit results, goodness of fit, and error analysis:
     independent points in data    =       8.152
     number of variables in fit    =       3
     degrees of freedom in fit     =       5.152
     r-factor of fit               =       0.002700
     chi-square                   ! ; =    1267.097168
 ! ;    reduce! d chi-square     &nb! sp;      =     245.926361
     feffit found the following values for the variables:
        variable            best fit value    uncertainty  initial guess
       ept            =       10.758198        0.860494        4.000000
       dpt1           =        0.006462        0.002930        0.004200
       sigpt1     &nb! sp;   =     &nbs! p;  0.005114        0.000098        0.005100
     correlation between variables
       variable #1    variable #2         correlation
       ept            dpt1                   0.926524
     all other correlations are less than    0.25000
---------------------------------------------------------------------------------------------
 Because the correlation between E0 and effective bonding length is big, I checked the fit as set bonding and the rest of parameters is variables,  and then the result is no! t changed. That means EO still is big.

&! nbsp;

_______________________________________________ Ifeffit mailing list Ifeffit at millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
_______________________________________________ Ifeffit mailing list Ifeffit at millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://millenia.cars.aps.anl.gov/pipermail/ifeffit/attachments/20120114/378fd8fa/attachment.html>