Dear Leonid, On Apr 3, 2014, at 7:16, Leonid Solovyov <l_solov...@yahoo.com> wrote:
>Dear Luca, > >I'm glad to see your interest to the problem even after such a delay. > >>So in the end, your model for faulting as you describe: >>"A more general model [J Appl Cryst (2000) 338] is included in DDM:" was to use a trigonal >>cell with hexagonal axis to allow refining the anisotropic shift of the >>peaks caused by the planar defects? That is a nice trick, you can try to >>justify it by reasoning that one effect of the planar defect is to get >>same stacking sequence of the hexagonal (we all know that), but what about >>the FCC stacking now? > >I chose the trigonal setting since the presence of layered faults decreases the symmetry of the >material making it intermediate between cubic close-packed and hexagonal close-packed. This is >an approximation of the real complex structure, of course, but I find it applicable to most cases >of close-packed faulted materials I've dealt with. In some cases one has to choose more >sophisticated models in order to account for peculiar diffraction features due to correlations >between faults, see, for instance Fig. 2 in [O. Ersen, J. Parmentier, L. A. Solovyov, M. Drillon, >C. Pham-Huu, J. Werckmann, P. Schultz, Direct Observation of Stacking Faults and Pore >Connections in Ordered Cage-Type Mesoporous Silica FDU-12 by Electron Tomography. >J. Am. Chem. Soc. (2008) 16800] It is what I anticipated in my message, finally you admit you use an approximation; in Maud I use instead the "real complex structure", so the FCC with the planar defects, all of three types as they cause different modifications to the pattern, and I obtain the density of each one. So I don't see any advancement in your approach to justify your initial claim of a better model. > >>To make the audience aware, just changing the cell was not sufficient, you >>have to reproduce the intensities. So in the structure a couple of Cu, Zn >>was set in 0,0,0 and another in 0,0,1/3 and the occupancies refined (to >>values as 1.10392 for the first position and 0.39608 for the second, but >>look like the second is calculated from the first) to adjust the >>intensities and the density. > >Yes, the additional (0,0,1/3) position is included in the model to account for the partial >displacement of atoms from ideal CCP sites due to the faulting. It allows accounting for the >influence of faults on both the intensities and the peak broadening according to the methodology >described in J Appl Cryst (2000) 338. >>But actually this didn't work out completely >>as the resulting quantitative phase analysis is completely wrong. > >Why do you think that the QPA is wrong? Just because it differs from what you have from >Maud??? No, because I made the sample, this is only one of a serie, and I analysed it years ago not with Maud, even before I started writing Maud, I analysed also it by neutrons and I know it very well. You should do some test and verify that when you use a model not based on the correct crystallographic setting, it is not assured that the total scattering will keep its intensity (there is an old paper of Warren, if I remember well was by him about it, I will try to find it). It means that it will affect your scale factor and in the end your volumetric quantity. Not to mention again that the B factor for your alpha approximation is completely wrong. This also affect your results. > >>And where are the crystallite sizes? Planar defect densities? So what kind >>of results did you get from the material science point of view? > >I can't determine the crystallite size from this pattern since I don't have the instrumental >broadening information. As for the faulting probability, it can be derived form eq. 5 of J Appl >Cryst (2000) 338. For the alpha-brass phase, the fraction C of defective cells (atoms displaced by >faults) is given by the total occupancy of the pseudo-position at (0,0,1/3) that is refined to 0.132. >The reciprocal values (1/t1 - 1/t) are listed in the column hkl of the DDM-output reflections >listing. For the faulting direction [003] this value is 64.2A. Thus, the fault probability: >p = 2Cd(001)(1/t1 - 1/t) = 2*0.132*2.1464/64.2 = 0.0088 >This value, however, may be biased as I don't have the instrumental broadening parameters. The instrumental broadening was in the Maud analysis and parameter file you used. You could have used it. Then you get one planar defect quantity representing both the two type of deformation faulting (slip) and twin. How you can mix twin and slip if hard to me to understand. >I >must also note that for such low-quality data one can hardly expect highly-reliable >microstructural characteristics. Well, in Maud i can, it is sufficient using the correct model. But it is not you who claim that the ddm model can do anything and magically get quantities out of nothing or I misunderstood your discussion with Armel? You have to relax a bit, the ddm is an interesting approach. I personally think is an improvement when dealing with medium/high crystallised samples. I am personally thinking to add it to Maud. But it is not the one for all solution. I don't think it is the best approach for nanosized samples as depending on the setting it may confuse the background/diffuse part with the large broadening oscillation of the nanophases, but for highly crystallised samples it take care of complex backgrounds in a clean way. But don't stretch it over everything, there is not one for all models, we have limits we cannot overcome. > >Best regards, >Leonid My best, Luca > >******************************************************* >Leonid A. Solovyov >Institute of Chemistry and Chemical Technology >660036, Akademgorodok 50/24, Krasnoyarsk, Russia >http://sites.google.com/site/solovyovleonid >******************************************************* >
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Please do NOT attach files to the whole list <alan.he...@neutronoptics.com> Send commands to <lists...@ill.fr> eg: HELP as the subject with no body text The Rietveld_L list archive is on http://www.mail-archive.com/rietveld_l@ill.fr/ ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
