Characterization of Standard Reference Materials Using Synchrotron Radiation Diffraction Data

Transcription

1 Copyright (C) JCPDS International Centre for Diffraction Data DENVER X-RAY CONFERENCE ( 1997) Characterization of Standard Reference Materials Using Synchrotron Radiation Diffraction Data Brian O Connor, Arie van Riessen and Graeme Burton Materials Research Group, Department of Applied Physics Cur-tin University of Technology GPO Box U1987, Perth, WA, Australia 6845 and David Cookson and Richard Garrett Australian Nuclear Science and Technology Organisation, PMB 1, Menai, NSW, Australia 2234 Abstract The value of using synchrotron radiation diffraction (SRD) for characterising standard reference materials (SRMs), as a supplement to laboratory x-ray powder diffraction (XRD) analysis, has been demonstrated with the National Institute of Standards and Technology LaB6 material NIST SRM 660. Measurements were performed under in vacua conditions with the high-resolution BIGDIFF Debye- Scherrer instrument (radius = 573 mm) at the Photon Factory, Tsukuba, Japan using a wavelength of A and a capillary-mounted specimen (diameter = 0.5 mm). A diffraction pattern recorded with imaging plates in 15 minutes provided pattern resolution and trace-phase (< 1 Oh) detectability in SRM 660 which are clearly superior to results for Bragg-Brentano laboratory XRD data collected in 1 hour. Search/match analysis of the SRD data readily revealed the presence of three impurity phases - a- Al203; the binary metal oxide La203*MO [M = metal]; and l&alumina material of the type 1 lal203*lazo:+*mo which were probably produced during the preparation of the SRM. While there are clear indications of some contaminant peak features in the corresponding XRD pattern, these could not be used to unequivocally identify the three contaminant phases. Rietveld phase composition analysis with the SRD data readily provided values for these trace phases. Introduction The present study was undertaken to evaluate the potential of BIGDIFF, a new synchrotron radiation (SR) diffraction (SRD) instrument, for characterising standard reference materials (SRMs). BIGDIFF is a multi-purpose SR materials analysis system operating at the Australian National Beamline Facility (ANBF) on Beamline 20B at the Photon Factory, Tsukuba, Japan [l-3]. One of the operating configurations of the instrument involves Debye-Scherrer optics, with a diffractometer radius of 573 mm,

2 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 ICDD Website -

3 Copyright (C) JCPDS International Centre for Diffraction Data corresponding to a 28 scaling factor of lo/cm along the instrument circumference; capillary-mounted specimens; and recording of the difiaction pattern with imaging plates (II%). BIGDIFF can rapidly provide (<< 1 hour) SRD patterns in Debye- Scherrer mode which are considerably superior in quality to those routinely measured with state-of-the-art Bragg-Brentano x-ray diffraction (XRD) instruments [3,4]. The study described in this paper is a sequel to an evaluation of BIGDIFF for defining low-concentration phases in ceramic specimens [3] with particular reference to the characterization of trace-level phases (< 1 A) which cannot be readily seen using laboratory XRD. Part of that study involved collection of SRD data for the National Institute of Standards and Technology LaB6 material NIST SRM 660 [5] to assess the instrument resolution. The SRD data for LaB6 showed a set of weak, but well-defined, contaminant Bragg peaks some of which were identified in a preliminary search/match analysis. The present paper describes a more detailed study of the LaBs data conducted to assess the potential value of using BIGDIFF-type optics with SR radiation for characterising SRMs - either by agencies, such as NIST, involved with SRM production or by users of SRMs produced by other laboratories. Assessment of the results of this paper should take into account recent improvements in the quality of laboratory XRD data for SRM analysis which have been achieved at NIST using parallel-beam geometry [6]. Experimental Trace element analysis of an SRM 660 specimen was conducted by the service laboratory SGS Australia Ltd (Perth, Australia) using an ICP procedure. For SRD data acquisition, a Si (111) channel-cut monochromator was used to provide a wavelength close to that of the CuKa doublet. Measurement of the wavelength with the NIST SRM 640 Si powder diffraction pattern gave a value of A. A slit combination placed in front of the monochromator controlled the length and width of the capillary exposed to SR. The incident beam dimensions were set at a height of 0.8 mm to completely immerse the specimen in the SR beam, and a width of 1.0 cm. An SRD pattern was measured with an as-received specimen packed in a low-absorption lithium borate glass capillary (0.5 mm diameter, 0.01 mm wall thickness). The specimen was rotated during data collection and the exposure time was 15 minutes. The data were measured with the instrument volume held at a pressure below 1 torr which largely eliminated air scatter attenuation at the wavelength employed. Four 400 x 200 mm erasable IPs were used, with the long dimension of the plate being parallel to the BIGDIFF circumference so that each plate spanned 40 in 28. The diffkaction pattern latent image on the plate was extracted in digital form by scanning with a Fuji BAS 2000 IP instrument along a strip extending 4 mm on both sides of the incident beam. The scanner employs a 20 mw He-Ne laser to release optical radiation which is recorded with a focusing optical system incorporating a photomultiplier tube (PMT). The output from the PMT was logarithmically amplified and digitised to produce an 8-bit image. The IP nominal pixel size of 100 pm converts to an angular resolution 0.01 in 20 which is substantially less than the ultimate resolution determined by the footprint of the capillary (0.05 for a capillary diameter of 0.5 mm).

4 Copyright (C) JCPDS International Centre for Diffraction Data Laboratory XRD data were measured for a LaB, flat-plate specimen at Curtin University with a Siemens D500 Bragg-Brentano diffractometer configured as follows - Cu tube operating at 40 kv and 30 ma (weighted mean wavelength = A), incident beam divergence = lo, receiving slit = 0.15, post-diffraction graphite analyser, NaI detector with pulse discrimination, 28 step size = 0.04, counting time = 1 s/step, 26 range = S-150 and pattern acquisition time = 60 minutes. Analysis of the Bragg peak profiles for the SRD and XRD data sets was performed with the program SHADOW (PC version 3.23, Materials Data Inc., Livermore, CA) as described in reference 3. The peak shape parameters were extracted from the data by profile refinement using a pseudo-voigt function which provided the FWHM values. Search/match identification analysis was performed with the Materials Data Inc program MICRO-ID Plus (PC Version 2.0) and the International Centre for Diffraction Data (ICDD) PDF-2 Database (Sets l-44). Identification of the trace phases was assisted considerably by studying the FWHM data - see Results and Discussion section. Phase composition analysis of specimen was performed by the Rietveld method with the SRD data using the Hill-Howard-Hunter LHPM program which accommodates multi-histogram data such as the set of four IP plates used with BIGDIFF [7]. Crystal structure data for LaBs and for the identified trace phases (atomic coordinates and unit cell parameters) were taken from the Inorganic Crystal Structure Data Base (FachInformationsZentrum and Gmelin Institut, Germany) - ICSD. The final Rietveld scale factors were converted to relative phase composition weight percent values using the ZMV expression, [s (ZMV)I analyte phase wt % (analyte phase) =, PI z Is ( ml all phases where s, Z, M and V signify the phase scale factors, number of formula units/cell, formula weights and unit cell volumes, respectively. The computations involved adjustment of the scale factors (phase and inter-plate), pattern-background polynomial function parameters, 20-scale offset and peak profile functions (pseudo- Voigt). Structural parameters were not refined. Attenuation was factored into the SRD calculations, as described in reference 3, using a specimen linear attenuation coefficient of 136 cm-. Results and Discussion Table 1 presents the trace element results. Figure la shows the intensity-versus-28 plot for the first IP of the SRD pattern (28 = O-40 ). The corresponding XRD data set is displayed in figure lb. While indications of impurity features were expected in view of the statement in the certificate of analysis for SRM 660 that a trace crystalline impurity was detected in the final powder [S], the corresponding XRD data for LaB, shown in figure lb clearly demonstrate the superior quality of the SRD data for low-concentration phases. The

5 Copyright (C) JCPDS International Centre for Diffraction Data 1999 Ms Al K Ca Cr Mn Concentration (ppm) l-2 Element Concentration (ppm) Fe 377 co* 10 Ni 16 cu 5 Zn 11 * By flame AAS The angular dependence of the FWHMs for the LaB6 peaks has been described in reference 3 using the model for a cylindrical specimen and a near-parallel beam emerging fi-om a double-crystal monochromator outlined by Cox et al l-81. The superior definition of the SRD pattern over that for XRD is evident from the statistic that the FWHM for the XRD data at 20 = 10 is approximately 0.15 compared with the SRD FWHM = The reduction in FWHM for the SRD data represents a factor x Figure 2 shows a plot of FWHM-versus-28 for the near-background peaks for IPl. It is evident from the data that there are four distinct sub-plots caused, presumably, by the development of different levels of non-linear strain in the three trace phases in excess of the strain present in SRM 660. The relatively low strain level evident from the LaB6 line widths is consistent with the SRM 660 Certificate of Analysis which states that there is a small residual microstrain line broadening observable at high 20 angles (typically 0.01 O above 1 lo ),,. The SRD search/match analysis results are illustrated in Figure 1 and summarised in Table 2. The analysis clearly showed the presence of three impurity phases (i) aalto - p eak numbers 8, 13 and 15 for plate IPl; (ii) the binary metal oxide La203*M0 [M = metal] - peak numbers 7, 11, 17; and (iii) a ternary l&alumina phase of the type llal,o,.la,o,.mo - peak numbers 4, 5, 10, 12, 14 and 16. The other near-background lines for IPl which have not been assigned (lines 1, 2, 3, 6, 9 and 18) point to the presence of at least one other trace-level contaminant phase. There is some ambiguity about the nature of the metal for the binary and ternary oxides due to similarities in the ICDD patterns for several metal types. None of the impurity phase lines could be attributed to W3 contamination from the monochromator, ie. if all d- spacings represented by the positions of the LaB, phase lines were reduced to one third, the corresponding pattern would not account for the faint impurity phase lines.

6 Copyright (C) JCPDS-International Centre for Diffraction Data The impurity phases appear to arise from the ball milling of the SRM material during preparation by NIST - refer to statement from SRM 660 Certificate of Analysis High purity LaB6 was ball milled, then annealed in Ar to reduce strain broadening. Agglomerates introduced by annealing were broken down by a short ball milling step Spectrochemical analysis prior to milling, etc indicated the material to be >99% pure. It appears that the impurities are due to the shedding of material by the milling media and mill container, and subsequent mechanical alloying interactions go A: CFAJO, [ B: [La,Ca]20,.Coz0, [ T: 2MgO. 11A~0,.La20, [ i,,,,, I,,,,,,,,,,,,,,,,,,,,,,,,,,,, #,,,, #,,, Figure 1. Low-angle portions of the diffraction patterns for the SRM 660 specimen: (a - top) SRD plot for h = A; and (b - below) XRD plot for h = A, CuKcz. The line assignments shown for the SRD pattern were identified by search/match analysis taking into account the FWHM-versus-28 plots in Figure 2 - symbols A, B and T refer to the phases a-alz03, La203*M0 [M = metal]; and 11Alz03*La203~M0, respectively. Symbol W in the XRD diagram refers to features caused by tungsten contamination of the x-ray tube anode.

7 Copyright (C) JCPDS-International Centre for Diffraction Data / 2MgO.l lal,o, ("I Figure 2. Plot of FWHM-versus-28 for the near-background SRD Bragg peaks. The large filled circles represent lines not assigned by search/match analysis. Table 2. Search/Match Analysis Trace-Phase Results for SRD Pattern Best-matched Phases ICDD - JCPDS # CCAl~O~ (La,Ca)zO&o203 [cubic] MgO.l 1Al,03.La20x The Rietveld phase composition results are given in Table 3. The crystal structure models employed in the calculations for the four phases were taken from the ICSD data base - the structures in references 9-12 for La&, a-al~03, (La,Ca)203.Co203 and 2Mg0.11AlzOpLa20,, respectively. It should be noted in considering these results that the wt % values are relative concentrations and that there may be at least one other phase present, albeit with wt % less than 1 %. The phase composition for LaBs agrees closely with the value 99 % specified on the SRM 660 Certificate of Analysis. The superior sensitivity of the BIGDIFF SRD analysis is evident from the concentrations of the trace phases which have been quantified at levels below 1 %.

8 Copyright (C) JCPDS-International Centre for Diffraction Data Table 3. Rietveld Phase Composition Analysis Results Phase ICSD* wt % Reference (rel) La (0.2)** CX-Al*O~ (0.30) La203.Co203 [cubic] (0.05) 2Mg0.1 1A1203.LaZ (0.22) * Inorganic Crystal Structure Database (Gmelin Institut) * * Values in parentheses represent the estimated standard deviations in terms of the least significant figures to the left. Conclusion The results underline the attraction of employing SRD for characterizing SRMs. The technique is clearly superior to Bragg-Brentano XRD in terms of the detection of trace phases and the assessment of line profile character, e.g. for residual strain analysis. SRD data collected with the BIGDIFF Debye-Scherrer instrument in only 15 minutes are clearly superior to Bragg-Brentano laboratory XRD data measured under typical data measurement conditions (approximately 1 hour acquisition time). The ready detectability of three impurity phases in the NIST SRM 660 LaB6 specimen, present at levels below 1 % concentration, in contrast with the XRD data set for which these phases could not be unequivocally identified, has shown that the detection limits for trace phases (cl%) in SRM analysis will be substantially superior with BIGDIFF SRD data. The result is clearly due to the superior dynamic range and angular resolution of the SRD instrument. Acknowledgement The authors wish to acknowledge a grant in 1994 from the Australian National Beamline Facility which is funded by a consortium comprising the Australian Research Council: the Department of Industry, Technology and Regional Development; the Australian Nuclear Science and Technology Organisation; the Australian National University and the University of New South Wales. We are also grateful to our colleague, D Y Li, for collecting the XRD data.

9 Copyright (C) JCPDS-International Centre for Diffraction Data References lr.f. Garrett, D.J. Cookson, G. J. Foran, T.M. Sabine, B. J. Kennedy and S. W. Wilkins, Powder Diffraction Using Imaging Plates at the Australian National Beamline Facility at the Photon Factory, Rev. Sci. Inst. 66, (1995). 2T.M. Sabine, B. J. Kennedy, R.F. Garrett, G. J. Foran and D.J. Cookson, The Performance of the Australian Powder Diffi-actometer at the Photon Factory, Japan, J Appl. Crystallogr. 28, (1995). 3B.H. O Connor, A. van Riessen, J. Carter, G.R. Burton, R.F. Garrett and D. J. Cookson, Characterization of Ceramic Materials with the BIGDIFF Synchrotron Radiation Debye-Scherrer Diffractometer, J. Amer. Ceramic Sot. 80, (1997). 4B.A. Latella and B.H. O Connor, Detection of Minor Crystalline Phases in Alumina Ceramics Using Synchrotron Radiation Diffkaction Data, J. Amer. Ceramic sot. 80, 294 l-2944 (1997). 5National Institute of Standards & Technology, Certificate of Analysis, Standard Reference Material 660. Instrument Line Position and ProJie Shape Standard for X- ray Powder Diffraction. Gaithersberg, Md. (1989). 6R.D. Deslattes, J.P. Cline, J.-L. Staudenmann, E.G. Kessler, Jr., L.T. Hudson and A. Hennins, Status of the Development of SRM 64Oc, 46th Annual Denver X-ray Conference (1997). Abstracts Volume, ~70. R. J. Hill, C. J. Howard and B.A. Hunter, A Computer Program for Rietveld Analysis of Fixed Wavelength x-ray and Neutron Powder Diffraction Patterns, Australian Atomic Energy Commission (now ANSTO). Rept. No. M112, Lucas Heights Research Laboratories, New South Wales, Australia, sd. E. Cox, B. H. Toby and M. M. Eddy, Acquisition of Powder Diffraction Data with Synchrotron Radiation, Aust. J. Phys. 41, (1988). gm.m. Korsukova, V.N. Gurin, T. Lundstroem and L-E. Tergenius The Structure of High-temperature Solution-grown LaBa: A Single-crystal Difiactometry Study, J Less-CommonMetals 117, (1986). 1eE.N. Maslen, V.A. Streltsov, N.R. Streltsova, N. Ishizawa and Y. Satow, Synchrotron X-ray Study of the Electron Density in a-al203, Acta Crystallographica B49, (1993). 11A. Wold and R. Ward, Perovskite-Type Oxides of Cobalt, Chromium and Vanadium with Some Rare Earth Elements, J. Am. Chem. Sot. 76, (1954). l2r. Brandt and H. Mueller-Buschbaum, Ein Beitrag zur Kristallchemie der Lanthanoidmagnetoplumbite, Zeit. fuer Anorg. und Allgemeine Chemie. 510, (1984).