RE:

[Peter Zavalij]

Dear Ling Yang, 1) Most of the patterns show texture (preferred orientation). It has to be refined if present. 2) Different patterns may have different texture. 3) Uiso should not be refined before and w/o texture and other correction, e.g. absorption/porosity (gsas) 4) That may be peak shape problem. Also extremely sharp peaks are present in case when the procedure of preparing powder (grinding) was not proper and some relatively large crystals are present. 5) extra peaks have to be identified, e.g. check if they match any of the cell that can form during transformation.   Peter Zavalij    ----- Original Message ----- From: Peter Zavalij To: [EMAIL PROTECTED] Sent: Friday, March 26, 2004 12:57 PM Subject: RE: This is quite comprehensive question(s) and even not so comprehensive answer could not be placed in one e-mail message. However, all these issues and more are covered in book "Fundamentals of Powder Diffraction and Structural Characterization of Materials" by V.K.Pecharsky & P.Y.Zavalij. Kluwer Academic Publishers, Spring 2003. (www.amazon.com, www.wkap.nl)   Peter Zavalij -----Original Message----- From: ling yang [mailto:[EMAIL PROTECTED] Sent: Friday, March 26, 2004 11:58 AM To: [EMAIL PROTECTED] Subject: Dear all,   I have some questions concerning the refinement of the gppd data and would appreciate your advice: 1) The patterns show some kind of texture(up and down on the error curve), but the sample is powder(although maybe the granules are large) and should be isotropic. All banks show this characteristic texture behavior. 2) For different banks, the sign of error for the same (hkl) gradually reverse, and it's hard to refine. 3) If UISO factors are let free, the refinement will go bad, even if they are constrained. 4) For one series of sample, the peaks are really sharp, the error sign is always like -/+/-, the positive error is very big and sharp. 5) There are some small peaks, which may be due to martensitic transformation. But they are really small, i don't know to what degree i should trust these data. Somebody said maybe that's due to contamination. If it is martensitic trans., how can i set a proper model to refine the data?     Sincerely,   Ling Yang PhD student in Dr. Wang's group 701 Scarboro Rd Oak Ridge, TN 37830 Tel: 865 574 0337 Fax: 865 241 5177