This is probably one of the most difficult things to achieve in XRD, and if you are a beginner, you have a tough journey ahead. Try Google and 'Structure determination from powder diffraction data'. This will give you some idea of what is involved. There is book on this subject - in the link below. A second link to an old review paper. A third link to another review paper.

The main problem is that while single-crystal data are 3D, powder data are only 1D, so information is reduced and ambiguity is increased.

The steps involved are - very roughly - Read as much as you can on structural crystallography before you begin, you will need to understand symmetry and space groups. Gather the very best data that you can; spend a long time on this, as rubbish data will give you rubbish results. Accurately locate the peak positions. From the lower angle positions use various auto-indexing software programs to search for a potential unit cell and a possible space group; look for consensus and a sensible physical solution given the number of atoms in your formula. With all your data, up to very high 2-theta values, use other software to attempt a structural solution; this is usually done by first building a guesswork trial structure and seeing if your data will refine onto it by various methods e.g. simulated annealing or charge flipping. You may need to trial several of the potential unit cells and space group combinations. Once a realistic model is found you can then start to use Rietveld refinement on that structure in order to more accurately locate the positions of the atoms within the unit cell. 

simples...

http://www.oxfordscholarship.com/view/10.1093/acprof:oso/9780199205530.001.0001/acprof-9780199205530

http://pubs.acs.org/doi/pdf/10.1021/cm960218d

http://www.unige.ch/societes/SSCr/w3cerny/Zkrist_1082-1_reprint.pdf

 This is probably one of the most difficult things to achieve in XRD, and if you are a beginner, you have a tough journey ahead. Try Google and 'Structure determination from powder diffraction data'. This will give you some idea of what is involved. There is book on this subject - in the link below. A second link to an old review paper. A third link to another review paper.

The main problem is that while single-crystal data are 3D, powder data are only 1D, so information is reduced and ambiguity is increased.

The steps involved are - very roughly - Read as much as you can on structural crystallography before you begin, you will need to understand symmetry and space groups. Gather the very best data that you can; spend a long time on this, as rubbish data will give you rubbish results. Accurately locate the peak positions. From the lower angle positions use various auto-indexing software programs to search for a potential unit cell and a possible space group; look for consensus and a sensible physical solution given the number of atoms in your formula. With all your data, up to very high 2-theta values, use other software to attempt a structural solution; this is usually done by first building a guesswork trial structure and seeing if your data will refine onto it by various methods e.g. simulated annealing or charge flipping. You may need to trial several of the potential unit cells and space group combinations. Once a realistic model is found you can then start to use Rietveld refinement on that structure in order to more accurately locate the positions of the atoms within the unit cell. 

simples...

http://www.oxfordscholarship.com/view/10.1093/acprof:oso/9780199205530.001.0001/acprof-9780199205530

http://pubs.acs.org/doi/pdf/10.1021/cm960218d

http://www.unige.ch/societes/SSCr/w3cerny/Zkrist_1082-1_reprint.pdf

Hi Mehri,

On youtube there are number of tutorial videos of Fullprof software, by which you can refine your xrd data. There is another software called Match, which will also help you to find the phase or structure that best matches with your xrd data. Can you share the compound name which you are trying to prepare or tell the starting materials and the expected final product name.

Hi,

i hope the newly developed compound (polycrystalline) may depend on your parent compound. for example if you take bismuth ferrite, it belongs to rhombohedral structure(R3c). however, it is add "X" compound in either "Bi" nor "Fe" site, then we may get an new compound that may or maynot be in the same phase. but still it belongs to the parent family and hence i try to find suitable matching  file using JCPDS or freely availalbe databases, based on the percentage of adding compound, try to found their phases too. after the matching statergy, we will try to resolve the phase fraction intensities with the help of availalble Rietveld techniques (GSAS,FULLPROF,MAUD, etc.,).