For example, if Sb3+ is doped to Fe2O3, and EXAF data is collected at Sb edge. then I want to fit Fe2O3 model to Sb3+ doped Fe2O3. Is this a right move?
when starting from the Sb k-edge you see the neighbours of Sb atoms. From these measurements you can estimate the Sb-O bond length(s?) and the SB-Fe bond.
But I think you want to have a look at the Fe-O bonds when you write that you will model the Fe2O3 (in the doped state). From the Sb as center atom it is difficult or even impossilble (to my opinion) to access the Fe-O bonds or a Fe2O3 model. But you may try it in any way.
But you may start modelling the Sb vicinity and then model F2O3 around (but I would no know how to do that, sorry).
For Fe-O it is better to start from the Fe k-edge. Here you have direct access to the Fe-O bond(s) and the Fe-Fe bond. However any difference to the undoped case will be very small due to the low concentration of the dopand. So also here a model for the Fe2O3 will not be easily achievable.
I collected data at Sb K edge and took Fe2O3 CIF input file for making cluster in atom and feff input file, assumed Sb /Fe are not distinguishable after doping. the energy edge was not considered, while EXAFS fitting on Sb K edge, using Fe2O3 CIF input.