Reliable quantification of Portland cement is a challenging task. The first step is to quantify the anhydrous cement first. Once you are confident to have a reasonable quantification routine, try to constrain or restrain the meaningful parameters in order to stabilize the refinement. Be aware that if you stop the hydration process you may destroy hydrate phases. Be also aware that the standard Rietveld quantification won´t give you the true phase contend of your cement paste, you will need to use the internal or external standard method. Adding an internal standard may however influence the hydration kinetics.

I don´t know why the link is not showing up:

https://www.researchgate.net/publication/229215693_Application_of_the_Rietveld_Method_to_the_Analysis_of_Anhydrous_Cement

Rule of thumb:

Main components: scale/Size/Strain/Lattice parameters (best from KOH treatment)

Sulfates / Alluminate / Ferrite: scale/Size/Lattice parameters (best from SAM treatment) 

Minor phases : scale/perhaps size  

Article Application of the Rietveld Method to the Analysis of Anhydrous Cement

It is difficult to tell what is wrong without looking at your results. Low R values is not the only thing to look at. In fact, it is very easy to obtain low R: it is enough to use the low-quality experimental data.

The most typical reason for unstable results is not the order of refinement (in all cases, your last step should be the refinement of all parameters together), but the set of parameters itself. Try to answer the following questions:

1. Are ALL important parameters refined? In other words, did not you miss something?

2. Are all parameters that you refine REALLY important? In other words, did not you refine something that in fact does not affect the quality of the fit?

2a. Do you have any correlations between the parameters? It is possible that by increasing one value and decreasing another one you get exactly the same fit. In this case obviously you can refine only one of two and another one should be fixed.

3. Is the difference between different runs really meaningful? How large is the variation of the parameters compared to the standard deviations?

In general, for any Rietveld refinement, the best is to start modeling the phase that shows the largest contribution to the diffraction pattern. Then continue with the phases  (one by one) with lower contributions to the diffraction pattern. It is important that the parameters of the "adjusted phases" continue free to continue their refinement.

Thanks for your responses Alexander / Xim,

Giving a lower and upper limit of the crystallite sizes of all the phases helped a lot in fixing the problem. I calculated approximate crystal sizes with the help of Diffrac Eva and allowed for +/- 15 to 20% upper and lower limits. Previously the at every refinement step the crystallite sizes of some of the phases used to get very wild values, which I was repeatedly resetting to a lower value in the next refinement. That was probably the reason, the results were varying wildly

This behaviour means that you have the typical problem of strong correlations between the profile parameters. As far as I remember, in TOPAS they are separated into instrument parameters, crystal size and strain. The instrument parameters should be calculated using some standard material and never refined. Size/strain parameters can be refined only for the major phases and should be fixed for the minor ones. Also, in order to refine all size and strain parameters you need to have data for as high angles as possible. If your measurement was stopped before 100 degrees, you have no chance to separate the size and strain contributions. In this case refine only one value for each phase and keep in mind that it will be purely mathematical parameter without physical meaning.

Reliable quantification of Portland cement is a challenging task. The first step is to quantify the anhydrous cement first. Once you are confident to have a reasonable quantification routine, try to constrain or restrain the meaningful parameters in order to stabilize the refinement. Be aware that if you stop the hydration process you may destroy hydrate phases. Be also aware that the standard Rietveld quantification won´t give you the true phase contend of your cement paste, you will need to use the internal or external standard method. Adding an internal standard may however influence the hydration kinetics.

I don´t know why the link is not showing up:

https://www.researchgate.net/publication/229215693_Application_of_the_Rietveld_Method_to_the_Analysis_of_Anhydrous_Cement

Rule of thumb:

Main components: scale/Size/Strain/Lattice parameters (best from KOH treatment)

Sulfates / Alluminate / Ferrite: scale/Size/Lattice parameters (best from SAM treatment) 

Minor phases : scale/perhaps size  

Article Application of the Rietveld Method to the Analysis of Anhydrous Cement

Thanks  Alexander / Dominique / Jennifer for your valuable suggestions.