Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 DISORDERING APPLICATIONS POLYCRYSTALLINE TO RIETVELD MATERIALS: REFINEMENTS Aurelio Cabeza, Enrique R. Losilla, H. Silvia Martinez-Tapia, Sebastirin Bruque and Miguel A. G. Aranda* Departamento de Quimica Inorghica, Mhlaga, 290 71 M&laga, Spain Cristalografia y Mineralogia, Universidad de e:mail: [email protected] ABSTRACT Three in-house adaptations of previously reported sample preparation methods to collect reproducible powder diffraction data on Bragg-Brentano diffractometers are discussed. This underlined the implications of the sample preparation method on Rietveld refinement results. The studied methods are: Side Loading Holder (SLH), Disordering with Spherical Amorphous Particles (DSAP) and Tubular Aerosol Suspension Chamber (TASC). Two materials have been selected for this comparative study: Pb(H03PC 6H 5) 2 which is a layered organo-inorganic compound that displays a very high preferred orientation effect; and Na~,~Zr&nr.~(PO& which is a 3-D ionic conductor of the NASICON family that displays non-reproducible powder patterns depending on the sample loading. The implications of the refinements on precision and accuracy of the obtained parameters and the possibilities for multiphase quantitative Rietveld analyses are also discussed. INTRODUCTION The applications of powder diffraction’ have been boosted by the routine use of Rietveld analysis2 However, for fully exploiting the information in a powder diffraction pattern, it is vital to have randomly orientated microparticles. Several sample preparation methods to get this type of “reproducible” X-ray powder diffraction data were compiled.3 These methods, mainly developed to reduce preferred orientation and hence, to obtain comparable standard data, are always advisable for a number of other reasons. Firstly, very low intensity peaks, which are not observed when the sample is oriented, become evident when the sample is disordered helping a lot in the autoindexing step when characterising new materials. Secondly, it is also helpful when ab initio structure determinations have to be carried out as the resulting structure factors are much less biased. Although preferred orientation can be adequately modelled with current algorithms (i.e. March-Dollase correction4), these sample preparation 228 228 This document was presented at the Denver X-ray Conference (DXC) on Applications of X-ray Analysis. Sponsored by the International Centre for Diffraction Data (ICDD). This document is provided by ICDD in cooperation with the authors and presenters of the DXC for the express purpose of educating the scientific community. All copyrights for the document are retained by ICDD. Usage is restricted for the purposes of education and scientific research. DXC Website – www.dxcicdd.com ICDD Website - www.icdd.com Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 methods also help in the structural refinements by the Rietveld method as they increase the number of observations in the patterns (higher number of measured diffraction peaks) leading to more accurate structures. Finally, the use is also very important when the sample is inherently formed by microparticles aggregates, as the intensities of diffraction peaks usually shown systematic deviations from the expected values. Hence, disordering small particles of the sample is important to record a reproducible powder pattern which matches the calculated one from the crystal structure. This is very adequate to obtain accurate crystal structures and to carry out quantitative phase analysis (by the Rietveld method). Sample preparation methods to collect reproducible powder data depend on the diffraction technique used and the information required. Sample used to obtain accurate d-spacing values may be treated differently from those used to extract intensities. Here, we will discuss the preparation methods for collecting data in a conventional powder diffractometer (BraggBrentano, W28, geometry) optimised for Rietveld refinements. The more common methods are: (1) Side Loading Holder (SLH) where the sample holder has a cavity extended to an end of the holder. The sample is allowed to drift into the cavity of the holder which is vertically placed with a glass slide clamped to it.5 (2) Disordering with Spherical Amorphous rarticles (DSAP) where the sample is diluted and blended with spherical nanoparticles (i.e. finely ground amorphous silica gel) to prevent the sample from becoming very oriented.6 (3) Encapsulating by Spray Drying (ESD) where particles of the sample are encapsulated in small spheres of organic-based material produced by a spray drying process.7 This is a very effective method although quite time-consuming. (4) Tubular Aerosol &tspension Chamber (TASC) where a highly uniform non-oriented sample layer is obtained on a substrate (i.e. glass tibre filter). Aerosol particles of the sample are generated by convection within a fluidised bed of glass beads.* In this work, three in-house adaptations of previously reported methods (SLH, DSAP and TASC) will be discussed in relation to Rietveld refinements. Two materials have been selected for this comparative study. Pb(H03PC6H5)2 is a layered organo-inorganic hybrid compound which displays a very high preferred orientation effect. Its structure was refined by the Rietveld method using the structure of Ba(H03PC6H5)2 (determined from single crystal data) as starting model.’ NazsZr&n1.~(P04)3 is a 3-D ionic conductor of the NASICON family being a member of the Nai+XZr2.,InX(P04)3 (02 x I 1.SS) solid solution. lo This compound displays non- 229 229 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 reproducible powder patterns depending on the sample loading very likely due to the presence of non-randomly oriented aggregates of microparticles. EXPERIMENTAL Pb(H03PC&)2 SECTION was hydrothermally synthesised as previously reported.’ Naz.gZro.&r,8(P0& was synthesised by the ceramic method. lo Four powder patterns were recorded for each sample using “normal” pressing loading, SLH, DSAP and TASC methodologies. The powder patterns were collected in a Siemens D5000 powder diffractometer. The patterns for Pb(H03PCbH5)2 were recorded between 15 and 42” (26) with 0.03” step size and 10 s counting time. The patterns for Na2.sZro.21n1,s(P04)3were recorded between 80 and 125’ (20) with 0.03’ step size and 5 s counting time. The working conditions of the X-ray tube were 40 kV and 40 mA. The optic in this diffractometer can be summarised as follows: a system of primary Soller foils between the X-ray tube and the fixed aperture slit of 2 mm. One scattered-radiation slit of 2 mm just after the sample, followed by a system of secondary Soller foils and the detector slit of 0.2 mm. Final powder patterns for structural refinements were collected on the DSAP prepared sample for Pb(HGsPC&)2 and the TASC prepared sample for Na2,8Zro,2Inl,8(P04)3with wider angular range 10-125’ (28) and counting 16 s per step.971o Rietveld refinements. The powder patterns were refined by the Rietveld method using GSAS” suite of programs. A pseudo-Voigt peak shape function12 corrected for asymmetry at low angles13was used to describe the diffraction peaks. The common overall parameters: histogram scale factor, background coefficients, unit cell parameters, zero-shift error and pseudo-Voigt coefficients including the asymmetry parameters (S/L and H/L) were refined.13 The peak shape parameters used were a Gaussian component, GW, expressed in (0.01°)2 units and a Lorentzian component, LY, expressed in (0.01’) units (Table 1). Pressed The samples were sifted through a 100 urn sieve. Then, the fine powder was deposited on a horizontally arranged holder and pressed from the top to get a flat sample surface. SLH. The sieved samples were loaded in the holder as indicated in standard reports.5 No further modifications have been tried. 230 230 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 TAX 231 apparatus. Figure 1 is a photograph of the TASC chamber, which is a modification of the early reported system.’ The chamber itself consists of a gas buret (30 cm of length) with a small flat section at the upper body. The internal diameter of this chamber is 3.2 cm. The lower stopcock of the buret which is not removed acts as the inlet for the air stream during suspension. This stream is produced by an air compressor. At the upper cut end of the buret is situated a glass microfibre filter substrate [GF/C - 0 4.7 cm (Whatman)]. This substrate can retain all sample particles with size bigger than 1 urn. The glass filter is attached to a metacrylate drilled disk to allow gasses to pass. The holes of the polymer disk has 0 = 1 mm. This set (filter and disk) are joined to the funnel filter holder with clips. The filter holder may be connected to a vacuum pump (for high density materials). In order to avoid spattering of the sample on the buret walls during loading, the chamber has a small lateral opening which is used to direct the mixture of sample and glass beads into the tapered, well above the stopcock. Optimum operation of the TASC method requires the use of a fluidised bed of fine glass beads (0 z 2 mm) in the well occupying about l-2 cm3 of volume (= 10 g). Figure 1. Photograph of the Tubular Aerosol Suspension Chamber TASC operation conditions. The sample has to be mixed in a vessel with the glass beads and the microparticles of the samples must remain adhered to the beads. This mixture (sample and beads) is placed at the bottom of the chamber to produce the desired suspension. 231 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Pb(H0$‘C&)2 was mixed with the glass beads with no problems, 232 however, Na2.sZro.&rr.~(PO& could not be adhered so easily and it was necessary to mix the powder with a few drops of acetone to get a quite dry paste in order to “wet” the beads. The output pressure of the air compressor was about l-2 bars. The air flow is finely controlled (and optimised) by the gas stopcock situated at the bottom of the chamber. For very high density materials it may be necessary to also apply vacuum through the filter holder on the top of the chamber. However, we have not used vacuum in this study. Several attempts to disorder the samples indicated that as deposition time of the sample on the substrate decreases, a higher orientation is observed in the final sample layer. It can take about 10 minutes to produce a homogeneous layer of the sample on the substrate. The optimum sample layer depth depends on the absorption of the material and it is quite difficult to obtain. The produced sample layer depth depends on the sample density and particle size, gas flow (and eventually applied vacuum), and time. Usually, = 0.1 mm yields good results with apparent zero-shift errors of the order of 0.25 degree (on Cu Ka, Bragg-Brentano diffractometers) for samples containing heavy elements. If in these conditions (statically), the layer sample still presents a high degree of non-random orientation, then, hand shaking of the buret is advisable which results in more randomly oriented microparticles on the filter substrate. Part of the sample (the bigger particles/aggregates) remained at the bottom of the gas buret with the glass beads or at the walls. The amount of sample required for TASC method is variable depending on the particles sizes, but usually = 250 mg gives a good sample layer. DASH We have used Cab-o-Sil M-5 (from Fluka) instead of the “similar” amorphous silica gel previously used.6 Cab-o-Sil are commercial silica amorphous spherical nanoparticles of approximately 30 nm of diameter. The Cab-o-SiVsample ratio (with a high time of grinding = 15 min) must be optimised for a given material to minimise the preferred orientation. This was monitored by short patterns of a selected region where the relative intensities are checked. The evolution of the anomalously high intensity peaks is followed, and although, the end of the process is “a priori” not known, we stop when not firther decrease of intensity is observed. 15 % in weight of Cab-o-Sil was found as optimum value for Pb(H@PC6H5)2, and others disordering works in our lab indicated that ratios between 10 and 20 % of Cab-o-Sil (about 50 % in volume) y ield good results. 232 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 233 RESULTS and DISCUSSION Pb(H03PC6H& study. In Table 1 are shown the refined parameters for Pb(H03PCsHs)2 obtained from the Rietveld analyses using the same structural model’ to tit the low angle region of the patterns collected on the same diffractometer with exactly the same optical configuration and varying only the sample preparation method as given in the experimental section. Pb(H03PC6H5)2 crystallises in a monoclinic, C 2/c, layered structure (with the a-axis perpendicular to the layers) as microplates, hence, showing strong preferred orientation. This effect can be taken into account by the March-Dollase correction with a coefficient smaller than 1.O along the <l OO>direction. The refined parameters are given in Table 1. The effects of preferred orientation are dramatic in the pressed and SLH patterns as shown in Figure 2. Table 1. Final refined overall parameters and disagreement factors for Pb(HG3PC6H5)2 Rwp/% Rp/% RF I % alA b/A CIA PI” v I A3 Zero-sh. I ’ Pref. ori. coeff. GW LY SIL Pressed 14.76 12.09 5.07 31.7783(21) 5.5894(4) 8.2797(6) 101.858(4) 1439.3(2) -0.010(l) 0.546(2) 5.4(6) 16.8(9) 0.020 0.020 SLH 10.32 7.53 2.63 31.7799(14) 5.5925(3) 8.2840(4) 101.865(3) 1440.9(2) -0.01 l(1) 0.566(2) 2.8(4) 21.4(6) 0.026 0.013 DASP 3.83 2.83 0.85 31.8148(13) 5.5976(3) 8.2897(3) 101.873(2) 1444.7(2) 0.067( 1) 0.916(2) 27.2(8) 26.6(7) 0.038 0.036 TASC 7.97 6.11 1.24 31.7939(19) 5.5932(4) 8.2835(5) 101.859(3) 1441.6(2) 0.307( 1) 0.921(3) 5.9(7) 32.8(8) 0.022 0.029 Average 31.792(17) 5.593(3) 8.284(4) 101.864(7) 1442(2) It is important to point out that although the precisions in the refined overall parameters is very high, the accuracy in Rietveld refinements is poor as shown in Table 1. The errors (xo,,) in the average unit cell parameters for Pb(HG3PC&)2 are almost an order of magnitude higher than the standard deviations for a single refinement. Although the asymmetry parameters, S/L and H/L, theoretically only depend on the optic of the diffractometer, in fact they depend on the sample and even on the sample preparation method. The variations are not very large, but we find them to be significative. There are not standard deviations for the values reported in Table 1 because they were not refined in the final cycle. They were optimised as described previously. 13bIt should be also noted that the refined peak shape parameters are different for distinct loading sample patterns. This effect is even 233 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 234 more important when working with materials showing anisotropic peak broadening due to the sizes of the coherent diffracting domains or the presence of microstrains, or both effects. We have found that in these cases, the changes in the peak shape parameters are dramatic. By reducing the intensities of some peaks is possible to model the anisotropic peak broadening much better as there are many peaks from different diffracting planes. Pb,HWCBH5,2~PRESSED Lambda, 5408A. LS cycle 245 I I thrt I ObSdrnd LM Profiler / 1 I I I Pb,HWCSHSI2-SLH Lambda15406A. L-S@de 286 I I Hllf 1 CJbSdandoil Prohiei I / i w gz i -- I I,, @ 3 I 15 II.., I 20 I , ,, , I, /I /a. 11111,II II .11 ,.,,,,,,I,,II I 25 30 I 35 2.iheta deg XlDE 1 Pb,HOJPCBHS)?DSAP IdP1 !m A.L-S qcie211 , I I 5 I 20 I 25 -1e I 20 I 25 I 30 2.Theta,deg B L 35 40 XlOE / 1 Ii *-Theta, deg / 20 1 JO I 35 I , HlSl 1 Obld andOfI PmBkl I ,. * L I 15 2-Theta.deg Pb(HWC6”5,2~ TASC Lsmbdst5406A. L-Scyde 251 I I ’ Hlrt 1 ObSd aId01 Profler I I r t 3 I 40 1 25 , . I JO I,, . , ., ,&,, ,,,,/, 35 I 40 XlOE , i ,. , , ,,,,/* ,,,,. .‘,.I I 40 XlDE I Figure 2. Rietveld fits for pressed, SLH, DSAP and TASC, Pb(HOJPChH5)2 DSAP and TASC methods a very powerful method to disorder materials that show strong preferred orientation effects (Figure 2). Although, preferred orientation can be describedlmodelled using current algorithms (Figure 2) the loss of information in the patterns is very important. This leads to less accurate structures because the number of observations (measured diffracting planes) are lower although the precision is not much affected as the number of measured points is the same in the four cases. 234 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 235 As it was discussed in the introduction, the loss of information also affects to the autoindexing step and the ab inito structure solution step if applicable. In fact, we failed to solve this structure ab initio, from SLH data and it was straight forward to get the structure from direct method from DSAP and TASC data. Nw4rdm.@‘04)3 study. In Table 2 are shown the refined parameters for Na2.sZr0.&r.8(P04)3 obtained from the Rietveld analyses using the same structural model” to fit the high angle region of the patterns collected as discussed above for Pb(HO3PC6H5)2. Na2.~Zro.21n1,8(P04)3crystallises in the 3-D NASICON belongs to the Nai+,Zr~Jn,(PO& tic type structure. This compound series, and from the variation of the unit cell parameters and from the 31PMAS NMR study” it was concluded that it is a single phase sample. As it can be seen in Figure 3, the intensities of the diffraction peaks depend upon the loading method. The disagreement factors for Na2.~Zro.21ni.8(P04)3(Rw~ and Rr) are very high due to the poor statistic of the data evidenced by the high noise level. However, RF dropped by more than 2 % for the TASC prepared sampled and the flat difference curve indicates that the structural model is correct. For the other sample preparation procedures, there was quite significant deviations in the observed intensities. Table 2. Final refined overall parameters and disagreement factors for Na2.8ZrO.2In1.8(P04)3 Pressed SLH 19.32 Rwpl% Rp I % RF/% 20.28 16.16 8.26 ali 8.9483(3) 8.9477(3) CIA VI A3 Zero-sh. I ’ GW LY 22.3344(8) 1548.8(l) 0.005(4) 22.3310(8) 1548.4(l) -0.008(4) 3.5(1.2) 9.2(4) 2.4(9) 15.74 8.13 8.8(3) DASP TASC Average 20.02 15.70 18.08 14.20 5.60 8.9544(3) 8.950(3) 22.3492(8) 1551.9(l) 22.339(8) 1549.8(1.6) 7.59 8.9509(4) 22.3423(11) 1550.2(2) 0.095(6) W) 9.8(6) 0.277(4) 2.5(1.1) 10.4(4) It can be concluded that TASC is also quite useful to obtain powder patterns of samples which display non-reproducible diffraction data probably because the microparticles are aggregated and arranged in a non-random way. TASC allows to obtain a layer of very small microparticles arranged in a random way and covering most of the possible directions. This can not be achieved when many aggregates are present in the sample. 235 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Na?BlZfl2lnl 8)1PC4)3~pressed Lambdat 5406A. L-ScyclelOBi HI* 1 Obrd andLM Profiler / / , 1 I Na28,20 2hl *,,PMp. SLH LambdaI we A. ts cycle,066 I * 236 HIS, 1 Oh andom Prm I / I 1 / 12 XlOE2 I I -= g :: %e ‘5 I I 8 2.Theta.deg I 10 8 NO28,ni 21”l 8,,Pc4,3 DSAP Lambda1 5406A.L-Scycle,015 / I I, z d I 12 XlOE 2 5 I 9 IO 1, 2ma. dlD Hlrt 1 Obsd and08 Profiler I 8 I Na28(m2,“, 8jp34j3~TASC Lambda1sdceP L.SElClrIt63 I 1 “,$I I Obrdand01 Prom I I I 8 I~Thetaa.deg Figure 3. Rietveld fits for pressed, SLH, DSAP and TASC, Na2.8Zro,2Inl,*(P04)3 However, TASC method should be cautiously used for quantitative Rietveld analysis of multiphase samples. In the sample preparation stage, smaller and lighter particles can ascend quickly that bigger and heavier ones of other phases leading to an inhomogeneous sample layer in depth. If in the multiphase sample layer obtained by the TSAC method, the phases are partially “ordered’ in depth, the results obtained will be non-sense, mainly for phases with quite different absorption coefficients. Hence, DSAP method with a light amorphous material is much more suitable for quantitative Rietveld analyses of phases which display ordered powder patterns. We have not implemented in our laboratory the ESD method so far, hence, it has not been discussed. Our experience indicate that SLH method is easy and well-known but it is not appropriated to reduce preferred orientation for materials that display this effect markedly. However, DSAP and TASC methods allow to disorder the particles very efficiently. Moreover, TASC method is also very useful to obtain powder patterns of samples which display nonreproducible diffraction data. 236 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Copyright(C)JCPDS-International Centre for Diffraction Data 2000, Advances in X-ray Analysis, Vol.42 Acknowledgement. This work has been supported by the research grant MAT97-326-C4-4 237 of CICYT, Spain. We also thank NATO for fimding through the NATO CRG program # 95 1242. REFERENCES PI J.I. Langford and D. Lotier, Rep. Prog. Phys., 59 (1996) 131-234 PI H.M. Rietveld, J. Appl. Cryst., 2 (1969) 65-71 PI H.F. McMurdie, Diffraction, PI M.C. Morris, E.H. Evans, B. Paretzkin and W. Wong-Ng, Powder 1 (1986) 5 l-63 A. March, Zeits. Krist., 81 (1932) 285-297. W.A. Dollase, J. Appl. Cryst., 19 (1986) 267-272 PI H.F. McMurdie, Dijjkaction, Fl M.C. Morris, E.H. Evans, B. Paretzkin and W. Wong-Ng, Powder 1 (1986) 40-43 M.C. Morris, H.F. McMurdie, E.H. Evans, B. Paretzkin, J.H. de Groot and R.J. Newberry, Natl. Bur. Stand (‘US.) Monogr. 25, 14 (1977) l-5 VI PI PI S.T. Smith, R.L. Snyder and W.E. Brownell, Adv. X-ray Anal., 22 (1979) 77-88 B. L. Davis, Powder Diffraction, 1 (1986) 240-243 D.M. Poojary, B. Zhang, A. Cabeza, M.A.G. Aranda, S. Bruque and A. Clearfield, J. Mater. Chem., 6 (1996) 639-644 WI E.R. Losilla, M.A.G. Aranda, S. Bruque, J. Sanz, M.A. Paris and A.R. West, Chem. Mater. (1998) submitted. [ 1 I] A.C. Larson and R.B. von Dreele, Los Alamos National Lab. Rep. No. LA-UR-86748 (1994) [ 121 P. Thompson, D.E. Cox and J.B. Hastings, J. Appl. Cryst., 20 (1987) 79-83 [13] (a) L.W. Finger, D.E. Cox and A.P. Jephcoat, J. Appl. Cryst., 27 (1994) 892-900. (b) M.A.G. Aranda, E.R. Losilla, A. Cabeza and S. Bruque, J. Appl. Cryst., 31 (1998) 16-21 237
© Copyright 2026 Paperzz