[Ifeffit] Manganese oxide EXAFS modelin

Matt Newville newville at cars.uchicago.edu
Thu Jul 12 22:10:14 CDT 2012


Hi Ebrahim,

On Thu, 12 Jul 2012, Ebrahim Rezaei wrote:

> Hi everyone,
> I am working with bulk and
> nano-particles manganese oxides. Basically I have collected EXAFS of Mn K-edge
> for my reference material (Mn2O3, bixibyite) and also the
> catalyst (Mn supported on alumina). The catalyst shows formation of
> nanoparticles of bixibyite on alumina with size around 20 nm.
> My first question is about modeling
> the reference material in order to find So which will be used as a
> fixed parameter to find oxygen and manganese first and second shell CNs for the
> catalyst.  
> The difficulty with modeling the
> reference material arises due to the rather complex atomic coordinates of Mn2O3
> containing oxygen atoms at different but very close distances. Basically Mn2O3
> has two sites (25% and 75% occupancy for site 1 and 2 respectively) with the
> following coordinates:
>
>
> Site 1                                                               Site
> 2 
>
> Atom               R (?)               CN                  Atom               R (?)               CN
>
> O                     1.993               6                      O                     1.899               2
>
> Mn                   3.104               6                      O                     1.985               2
>
> Mn                   3.565               6                      O                     2.249               2
>
>                                                                         Mn                   3.115               6
>
>                                                                         Mn                   3.575               6
>
>  
>
> I am aware that in order to fit the
> EXAFS equation I should considers paths from both sites and multiply their
> amplitudes by their site occupancy. But to make the life easier I simplified
> the atom coordinates of Mn2O3 to be able to extract the
> CNs from the catalyst. To my understanding one can expect to have 6 oxygen
> atoms in average at the first shell (0.25*6 + 0.75*6 = 6), 6 Mn atoms at the
> second shell and 6 Mn atoms at the third shell regardless of the site occupancy.
> Therefore I ignored the site
> occupancy and considered the following atomic coordinates assuming that there
> is only one site in the in the model developed for the reference material:
>
>  
>
> Atom               R (?)               CN
>
> O                     1.985               4
>
> O                     2.249               2                                  
>
> Mn                   3.115               6                                  
>
> Mn                   3.565               6          
>
>  
>
> The model works fairly acceptable
> and I am able to get CN from the catalysts as well. But I am not sure if my
> assumptions could be acceptable or not. I was wondering if I can have idea of the
> EXAFS community experts on this.

I may have gotten a little lost here, as you started off with 
trying to get S02 from Mn2O3. If Mn2O3 is so structurally 
complicated (and, it is), perhaps it's not the best choice for 
getting S02. After all, the whole point of a reference is that 
you are assured of the structure that any apparent change in 
coordination number must be due to S02.

> My other question is about dealing
> with ?2 as a fixed parameter. I have noticed that after fitting the
> catalysts, ?2 of paths in the catalysts are almost close to that of
> the reference material. Therefore I tried to fix the values of ?2 for
> the catalyst and use the ones I have obtained from the reference material. This
> increases the degree of freedom of the fit and results in smaller error bars in
> CN while ?2 increases by 2.5%. Basically I have always kept number
> of variables to almost less than half of the number of the independent points
> either with fixing or relaxing ?2.
> I was wondering if estimation of ?2
> in the catalysts from the reference material might be a good assumption
> in my case or not. I think it should not be a bad idea since size of Mn2O3
> nano-particles on alumina is fairly large (around 20 nm) showing formation of
> complete crystalline of Mn2O3 on the catalyst.  

I'm not sure I follow your reasoning, but perhaps that's because 
I wouldn't necessarily assume that "large nano-particles" are
"complete crystalline".   And, if the reference material is as 
complex as Mn2O3, how you take into account the inherent 
structural disorder would seem to be important.

I don't necessarily disagree with you here, but I'd urge you 
to be generally cautious about transferring values for sigma^2 
between complex, multi-site metal oxides.   I would probably 
allow sigma2 to vary, and be happily unsurprised when the value 
does change much between different sample rather than try to 
make complex arguments about why it can be assumed to be the 
same for different samples.

--Matt


More information about the Ifeffit mailing list