> I just tried to compare four different software: FindIt, Diamond, > PowderCell and FullProf to calculate structure for alpha-alumina using > ICSD #51687 file (Materials Science Forum(2001), 378, 288-293,Unit > Cell 4.7597(1) 4.7597(1) 12.9935(3)). I used the same geometry, > wavelength, profile parameters, no background, same space group > settings. But all four patterns have different intensity, and structure > factors also (for software that gives a possibility to get one). I can > not understand why the structural factors differ. The visualized > structures are exactly the same. Maybe somebody knows?
Dear Irina, I do not know how big the differences are, so maybe some of the differences are not big enough to be relevant. However, trigonal space groups like R-3c are more difficult to get right than the more straightforward lower symmetry space groups. This is because the matrices representing the symmetry operators are no longer diagonal, which means that it becomes important to multiply them in the correct order (from the left or from the right). Not all programs get this right, which could explain the differences you are observing. Secondly, in trigonal space groups special positions can contain values like 1/3 and 2/3, which can no longer be represented accurately in text files but are truncated to values like 0.3333 and 0.6667, which can give problems with rounding errors; however, the atomic coordinates of corundum do not contain these values, so this cannot be the problem. And Vincent already mentioned another problem, namely the confusion over the hexagonal versus the rhombohedral setting. If the differences that you observe are substantial (i.e. big enough that at least one of them must clearly be wrong), then it is important to first establish what the correct answer is. I found an experimental powder pattern of corundum (CuKa) on the web: http://www.ccp14.ac.uk/tutorial/powderx/runthr.htm Using the coordinates from ICSD #73724, which can be found here: http://www.crystal.unito.it/mssc2006_cd/tutorials/geometry/geom_tut.html it can easily be tested which program gives the correct answer. I verified that the visualiser Mercury, freely available for a variety of platforms from http://www.ccdc.cam.ac.uk/products/csd_system/mercury/ gives the correct simulated powder pattern. The next version of Mercury, Mercury 1.5, due to be released in a couple of weeks, will also able to write out structure factors, and it will also have the option to include CuKa2. Kind regards, -- Dr Jacco van de Streek Send instant messages to your online friends http://uk.messenger.yahoo.com
