Hi Nirmalendu,

 what do you mean by 'EXAFS input path'?

From EXAFS measurement of such a phase mixture you will get peaks in the ampliute distribution of the  Fourier transform of the EXAFS modulation, the positions of which represent the distances of the scattering atoms to the center atom.  But I think due to the manifold phases  these  distances are very close together, so you will get a large overlap of the peaks. You will propably not be able to really separate all the peaks. 

Perhaps you want to fit the EXAFS structures of your 'mixed phases' sample and find out the relative contributions of the different phases.

My suggestions is to calculate the EXAFS of each phase, preferently matched to measurements of each single phase, and afterwards  performing a weighted sum of the EXAFS structures of your single phases. Now you have to fit the weighting factors of your 'calculated' phase mixture to the EXAFS of your 'measured' sample. You can do this comparison directly with the EXAFS structure or including the  Fourier transform.  

It's hard enough to constrain an EXAFS fit for a single unknown phase - four phases will be a real challenge! What specific questions are you trying to answer with EXAFS?

Here are some suggestions for your analysis:

• If some of the four structures are known, input paths can be generated from published atomic coordinates or .cif files. If not, you can construct a guess structure or duplicate selected paths from a known structure.
• In order to fit the EXAFS, you'll have to fix as many parameters as possible, guided by your XRD data. If the Scherrer broadening of the XRD peaks indicates sufficiently large crystallite sizes for the phases with known structures, then you may be able to constrain the bond lengths, coordination numbers, and sigma^2 values of those phases to published bulk values.
• If you have an accurate measurement of the fraction of each phase, you can constrain the ratios of the path amplitudes of each phase to those ratios, similar to Gerhard's suggestion.

Thank you Dr. Martin for your suggestions.

What is exactly the scientific question you want EXAFS to answer? Though you can attempt some components fitting, as suggested by Martens, it will only work if the crystallites in your sample have a high crystallinity and reasonable size, as pointed out by McBriarty. Otherwise the increased DebyeWaller and the amount of amorphous material will conspire against your fit.