Dear Armel, I have re-adjusted my tasks in order to start with the examination immediately. Could you, please, send me the exercise?. Thanks. Leopoldo
Armel Le Bail dijo: > Hi, > >>Few days back I have put a question on the list but has not yet replied >> me. >>I desperatly need your help. > > The fact is that your question has a complex answer... > >>I have evaluated the instrumental parameters U,V,W from cagliotti >> relation >>and 'a' and 'b' from the eta relation. >>Can anybody tell me is there any program/software which will use these >>informations of the instrument and deconvolute the instrumental profile >>from the xrd pattern and give me the error free diffraction profile. > > No, there is no serious program doing that, IMHO. > >>I am also interested to obtain the instrumental corrected intetnsity >> data. > > It is already difficult to deconvolute one single well defined peak > and to obtain the sample profile f (where the convolution is > h=f*g; h being the experimental data, g being the instrumental > contribution and f being the pure sample effect). So you need the > f contribution for the whole pattern ? Rather difficult... Some old > papers published at the Denver Conference pretended to be able > to do it but they were not really serious contributions (I can try > to remember the references but I would not recommend using > the method).. > > The parameters you call "the instrumental parameters U,V,W from > cagliotti relation and 'a' and 'b' from the eta relation", how were they > evaluated ? from a NIST standard material ? Which one, or something > else? With what radiation ? What is the chemical composition of the > sample to which you expect full transferability of these U, V, W, etc > values ? Do you realize that such a description of the experimental effect > is quite a bad approximation if it has to be used in a deconvolution > process > (which is very unstable) ? > > People prefer to convolute g (or a model of g) with a model of f in > order to obtain h, this is much more stable than to deconvolute. > Nobody works on the f full profile (you would be alone) because > nobody can obtain a realistic representation of it, without strange > effects (ripples, etc) due to error propagation (either by Fourier > inversion method or else). > > Or may be ask for help from a Fundamental Parameters approach > since the g contribution is theoretically completely defined with > some geometrical mounting (you even do not need a NIST reference > sample, or whatever), in principle better approximated by convoluted > profiles than fitted with only one profile shape, an angle-dependent width > (U, V, W) and an angle dependent shape. But even with that FP > approach, obtaining a realistic f full profile is not really feasible. > > However, I am old now, and things may have evolved ?-). Is someone > deconvoluting daily his whole powder pattern somewhere, removing > the instrumental contribution ? > > Best wishes, > > Armel > > PS - do not confuse deconvolution with desummation. > > > > > > > -- Dr. Leopoldo Suescun Pereyra CSSC-Lab./Dequifim Facultad de Qu�mica, Universidad de la Rep�blica Av. Gral. Flores 2124, C.C. 1157. 11800 Montevideo, Uruguay. Tel: (+598 2) 9290705 Fax: (+598 2) 9241906 URL: http://cssc-lab.fq.edu.uy/Leopoldo/leo.htm "No vender� el rico patrimonio de los orientales al bajo precio de la necesidad" Jos� Artigas
