That's right, doing series of samples brings great confidence that everything was done right (of course if there is agreement the series), especially when you are approaching border line of the possible, which may be the case here since 4-5% of transition metal is only 1 electron.
Peter Y. Zavalij Director, X-ray Crystallographic Laboratory Department of Chemistry & Biochemistry 091 Chemistry Building University of Maryland College Park, MD 20742-4454 Phone: (301)405-1861 Fax: (301)314-9121 E-mail: [EMAIL PROTECTED] -----Original Message----- From: Alan Coelho [mailto:[EMAIL PROTECTED] Sent: Thursday, May 05, 2005 2:39 PM To: rietveld_l@ill.fr Peter It sounds like you were able to refine on the occupancies without physically unreasonable site occupancies. I am also sure that you have simplified the discussion for clarity. This is fine but I would add as a hint to Alexander that making a number of series of samples can assist greatly in the analysis. On a single sample it is a difficult job of determining precise occupancies; this we all agree upon. For a series of samples, lets say 6, change a single variable, a dopant concentration for example. How, the various Rietveld parameters behave for the various phases involved with respect to the dopant concentration yields convincing proof of what is happening. The parameters to look at of course include lattice parameter, bond lengths, temperature parameters and the total scattering power observed on the sites. Thus if the scattering power on a particular site changes with dopant concentration then you can be sure that that site is being affected - a simple enough deduction to make. If the site in question has a similar scattering power as the dopant but different ionic radii then the change in the site occupancy should show up in the trend seen in the average bond lengths. Further trends are obtained with many other series of samples with sometimes only a few peaks being analysed rather than doing full refinements. Together with refinement and enough series of samles the trends and site occupancies can be exposed - maybe. Thus to end I would recommend looking for trends rather than just looking at a single sample and believing the refinement results. Alan -----Original Message----- From: Peter Zavalij [mailto:[EMAIL PROTECTED] Sent: Thursday, May 05, 2005 8:16 PM To: rietveld_l@ill.fr That's right. The assumption was made just to simplify discussion; in reality of course there were no assumptions. By the way in that particular case it is possible to refine more that 1 occupancy (Ni/Li), e.g. total metal content. Peter Peter Y. Zavalij Director, X-ray Crystallographic Laboratory Department of Chemistry & Biochemistry 091 Chemistry Building University of Maryland College Park, MD 20742-4454 Phone: (301)405-1861 Fax: (301)314-9121 E-mail: [EMAIL PROTECTED] -----Original Message----- From: Alan Coelho [mailto:[EMAIL PROTECTED] Sent: Thursday, May 05, 2005 12:14 PM To: rietveld_l@ill.fr Peter I dont know your chemical system but my first thought is that assuming full occupancy may be too much of an sssumption. In the system I looked at vacancies of up to 7% were observed. In fact the vacancy limit probably determines the stability of a phase and when the limit is reached there's a phase transition. It may very well be the case that your system has no vacancies but I would not assume it at first. alan -----Original Message----- From: Peter Zavalij [mailto:[EMAIL PROTECTED] Sent: Thursday, May 05, 2005 6:44 PM To: rietveld_l@ill.fr We didn't assume that Ni goes to the Li sites it was the only choice confirmed by others, and it is not so simple as the full occupancy and stoichiometry should be measured accurately. BTW sample composition as well as sample preparation is a very crucial factor where and what goes. I believe that Mn1/3Ni1/3Co1/3 is more complex than our composition. Peter Y. Zavalij Director, X-ray Crystallographic Laboratory Department of Chemistry & Biochemistry 091 Chemistry Building University of Maryland College Park, MD 20742-4454 Phone: (301)405-1861 Fax: (301)314-9121 E-mail: [EMAIL PROTECTED] -----Original Message----- From: Whitfield, Pamela [mailto:[EMAIL PROTECTED] Sent: Thursday, May 05, 2005 11:08 AM To: 'rietveld_l@ill.fr' Cc: Whitfield, Pamela Since we're on the subject of battery materials, we published some work recently where we didn't assume that it was the Ni that went to the Li site in LiMn1/3Ni1/3Co1/3O2. It's logical if it's Ni2+ but we had the data to test it in the form of resonant scattering data to add some more information to the puzzle. We could let the TMs (ratios known from XRF) float over whichever site they wanted together with refining the Li/TM ratio. Now if you have Li-rich Mn/Co/Ni materials where the Ni isn't necessarily in the +2 state then things get a bit more muddy as all of a sudden the TMs have similar ionic radii. :-) Monoclinic symmetry also makes it a lot more fun (?!) Pam Dr Pamela Whitfield CChem MRSC Energy Materials Group Institute for Chemical Process and Environmental Technology Building M12 National Research Council Canada 1200 Montreal Road Ottawa ON K1A 0R6 CANADA Tel: (613) 998 8462 Fax: (613) 991 2384 Email: <mailto:[EMAIL PROTECTED]> ICPET WWW: http://icpet-itpce.nrc-cnrc.gc.ca -----Original Message----- From: Peter Zavalij [mailto:[EMAIL PROTECTED] Sent: May 5, 2005 11:36 AM To: rietveld_l@ill.fr Alexander: I just finished combined X+N refinement of similar battery materials but w/o V with Li in 2a site and Mn, Co and Ni in 2b (or vise versa). Chemical composition was well known; The problem we were looking for was migration/exchange of transition metal from 2b to 2a. That's not so simple problem as basically we have more unknown parameters than equations, counting that trans. metals are practically undistinguishable from x-ray. Fortunately their neutron scattering coefficients are very different (Mn -.37, Co .25 & Ni 1.03). It happens that Ni migrates to the Li sites and since it has the largest Neutron scattering it is the only possibility. The logic is simple: 1) X-ray yields that ~ 6% of trans. Metal (but we don't know which) migrates to the Li sites. 2) Neutron gives the same 6% only if the migrating metal is Ni. Thus, in this case the only refined parameter is amount of Ni in Li site: Li(1-x)Ni(x) in 2a AND Mn(k) Co(m) Ni(n-x) Li(x) in 2b, where k,m,n are known. It assumes however that: 1) stoichiometry is well established and 2) both sites are fully occupied. The latter can be confirmed for example by measuring density. Good luck Peter Y. Zavalij Director, X-ray Crystallographic Laboratory Department of Chemistry & Biochemistry 091 Chemistry Building University of Maryland College Park, MD 20742-4454 Phone: (301)405-1861 Fax: (301)314-9121 E-mail: [EMAIL PROTECTED] -----Original Message----- From: Alan Coelho [mailto:[EMAIL PROTECTED] Sent: Wednesday, May 04, 2005 3:03 PM To: rietveld_l@ill.fr Alexander: Your problem is quite similar to one I had to solve in my thesis where I had a number of mixed valence sites. I also had at my disposal X-ray and neutron data. The method that I will now mention may be manipulated to help. I found the need to match expected stoichiometric results (known elemental composition) with weight percents obtained with Rietveld refinement including contributions from impurity phases. Care needs to be taken in regards to micro-absorption effects. The stoichiometry is known if you are synthesising the samples; if not then XRF results are necessary. The effect of using Stoichiometry is to add another 4 to 6 constraints. Then like you suggest you can form a number of equation constraints but some of them are not linear. Except for a scaling constant the scattering power amongst the various cations are too similar for X-rays and non-existent for neutrons. The most important factor is the synthesising of a series of samples with an expected vacancy concentration. In other word working from a single sample does not generally contain the contrast that you need. Only from relative changes in phase concentrations can you then determine the vacancy concentration of the target phase. It's quite involved and I can send you my thesis on request. A lesser use of the vacancy concentration process can be found in the publications: Cheary, R. W. & Coelho, A. A. (1997). "A Site Occupancy Analysis of Zirconolite CaZrxTi3-xO7". Phys Chem Minearls, 24, 447-454. Coelho, A. A., Cheary, R. W. & Smith, K. L. (1997). "Analysis and Structural Determination of Nd Substituted Zirconolite 4M". J. Solid State Chem., 129, 346-359. All the best Alan -----Original Message----- From: Alexander J.M. Schmets [mailto:[EMAIL PROTECTED] Sent: Wednesday, May 04, 2005 5:38 PM To: rietveld_l@ill.fr Dear users of the Rietveld mailing list, My name is Alexander Schmets and currently I work as a PhD student in the Neutron scattering department at the Delft University of Technology, The Netherlands. I read this Rietveld already quite some time, but this is my first question. 1) I have a range of samples containing Li, V , O and one or more other transition metal ions (Ni, Co, Mn, Fe). It seems beneficial to do a combined experiment: neutrons for finding the Li (V hardly visible), and to distinguish between the transition metal ions; X-rays to get the vanadium occupation correct. I have high quality X-ray as well as neutron diffraction (GEM) data. What now, is royal way to proceed, such that the 'contrast' is optimally benifitted from? (there is a topic already about simultaneous refinement on this list, though it couldn't help me too much). The structure is a mixed spinel (F D -3 m), where Li,V and the transition metals share the 8a, 16d sites and oxygens are as usual on the 32e sites. 2) I use GSAS to refine the structure. The transition metals can occur in a range of oxidation states (eg: V5+, V4+, V3+, V2+, V). Different oxidation states will contribute differently to the scattered x-ray intensity. At the same time V's in different oxidation states will have different 'bond lengths' with their coordinating oxygens. Consider I know (from other experiments) that V5+ (partly) occupies a 16d site ...should I attribute instead of V the element that is five places backwards (Argon) to that site, in order to have the correct scattered intensity? And then ... the bondlength definately goes wrong...should I fix it, and where to find an apropiate estimation for such bond length? May be too many questions for a first appearance on the list. But I don't see a way out. Best Regards, Alexander nb) I got already the following advise: put hydrogens on all lattice sites ..refine the fractional occupations of the sites ... then use a priori knowledge about which elements/oxidation states reside on these lattice sites ...and one has a set of linear equations to solve. This would give a set of possible structures that could be starting point for further refinement (with now fixed partial occupancies) -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* Alexander J.M. Schmets Departme
