USABILITY OF PORTABLE X-RAY SPECTROMETER FOR DISCRIMINATION OF VALENCE STATES

Transcription

1 Copyright (c)jcpds-international Centre for Diffraction Data 00, Advances in X-ray Analysis, Volume ISSN USABIITY OF POTABE X-AY SPECTOMETE FO DISCIMINATION OF VAENCE STATES I.A.Brytov,.I.Plotnikov,B.D.Kalinin, NPP Bourevestnik, Malookthinsky Pr., 68, St. Petersburg, ussia ABSTACT Spectral resolution and line positioning reproducibility of the SPAK-1M portable X-ray spectrometer have been determined and the usability of the instrument for the study of chemical shifts of K-lines in the emission X-ray spectra of transition metals has been evaluated.the spectrometer has been shown to reproduce goniometer positioning within a fraction of an angular second, which corresponds to the line shift on the order of 0.01 ev. In the case of iron, chromium and uranium compounds with different oxidation states, the usability of the spectrometer for measuring chemical shifts and for the assay of transition metal oxidation states has been demonstrated. INTODUCTION In certain analytical problems, a ratio of different oxidation states of a given chemical element has to be determined, such as the ratio of sulfate and sulfide in coal, Fe + и Fe +3 in kimberlites, mica and other minerals, UO + and U +4 in nuclear materials, etc. Precision X-ray spectrometers allow to solve these problems by studying the fine structure of X-ray emission spectra and measuring X-ray line shifts (chemical shift) [1,]. Usually, chemical shifts do not exceed ev and are difficult to measure. The use of conventional Soller-type scanning spectrometers (mass-produced for elemental analysis) to study chemical shifts has been reported [3,4]. This present paper reports on the use of a portable X-ray spectrometer to measure chemical shifts of Kβ-lines of transition metals and U α line. Measurement errors are calculated. SPECTOMETE CHAACTEISTICS SPAK-1M is an automatic short-wave Johansson-type X-ray spectrometer with a 150 mm focal radius. A BKh-7 X-ray tube capable operating with an anode voltage up to 45 kv and output power up to 10 W was used as an excitation source. The anode of BKh-7 tube is a thin silver coating (about 5 µm thick), located directly on the tube beryllium window. This design allows one to bring the sample within 3-5 mm from the focus and to obtain high count rates, which is only ten times lower than those on spectrometers with 3 kw reflectiveanode X-ray tubes. The scanning mechanism simultaneously rotates the crystal and the detector (a xenon-filled proportional counter) at the correct ratio within the range θ = 4 88 о. With the most common crystal, if(00), this range corresponds to the wavelengths from 0.83 to.8 Å, covering the K-series of elements in the range from Z=1 (Sc) to Z=51 (Sb) (using first and second orders of reflection) and the -series of elements with Z 56 (Ва). A stepper motor scans the spectrum in Å steps throughout the entire wavelength range. ine half-width was measured and calculated as a geometric sum of four components: intrinsic line width, entrance slit width (τ=0. mm), mosaic non-uniformity of the crystal (δ =3 ) and vertical divergence of the beam (crystal width of 0 mm).

2 ISSN 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 00, Advances in X-ray Analysis, Volume ISSN For wavelengths over 1.3 Å, full agreement is observed between calculated and experimental data (from 8 ev for TiKα to 67 ev for Zn Kβ). For shorter wave lengths (U α line in the second order), measured values significantly exceed the calculated ones (98 and 56 ev), which may be explained by the passage of radiation through the slit blades and penetration of the radiation into the crystal. eproducibility error of the goniometr was evaluated by repeated measurements of the count rates on the line slope with repositioning of the goniometer after each measurement. The average result of 10 goniometer repositioning cycles r.m.s. reproducibility was 0.01 ev or 0. angular seconds, which can compete with the reproducibility of precision spectrometers. Similar results were obtained for several individual SPAK-1M spectrometers. With the reproducibility and resolution demonstrated, line shifts due to change in chemical state of an element may be measured. X-AY SPECTA OF Cr Kβ, Fe Kβ, AND U α INES Chemical shifts was evaluated by measuring parameters of a differential spectrum (difference between normalized profiles of the lines in the compounds with different oxidation states of the given element), or by measuring the ratio of intensities on the left and right slopes of the line. The latter procedure substantially reduces measurement time and is especially convenient when the sample contains the given element in both oxidation states. In this case, the emission line is a superposition of two lines with different shift, and the shift of the measured line gives direct quantitative information on the ratio between the oxidation states. Figs. 1 and present the comparison of spectra of iron (III) oxide Fе O 3 vs. iron (II) oxalate FeC O 4 (Fig.1) and chromium (III) oxide Сr O 3 vs. potassium chromate K CrO 4 (Fig.) in the Kβ spectral region for iron and chromium, respectively.

4 Copyright (c)jcpds-international Centre for Diffraction Data 00, Advances in X-ray Analysis, Volume ISSN Fig. 3 presents the α1 line of uranium oxides UO and UO 3 in the second order of diffraction. These were measured the powder probes of all mentioned compounds. A voltage of 35 kv and current of 0.mA were used. Peak count rates were ~40,000 s -1 for iron, ~1,000 s -1 and ~4,000 s -1 for chromium in oxide and chromate, respectively, and ~4,000 s -1 for uranium. The spectra were scanned in Å steps with 10 second exposure at each point. For every sample, an average result of 5 measurement cycles was used.upon normalization of the spectra (the value at every point was divided by the sum of all points), differential count rates J 1 J were determined and plotted vs. wavelength together with the line profiles (differential curves for chromium and iron are magnified for a better view).

5 Copyright (c)jcpds-international Centre for Diffraction Data 00, Advances in X-ray Analysis, Volume ISSN As Fig. 1 shows, differential count rate J Fe+3 J Fe+ (dotted line) has a fairly complex trend; its S-shaped profile testifies to the shift of the main Fe Kβ 1,3 peak.a peak around Å and additional peaks superimposed onto the main curve around Å are due to the Fe K β,5 line whose intensity and position depend on the oxidation state of iron. A similar structure around longer wavelengths ( Å) is due to the non-diagram line Fe Kβ with the position and intensity also substantially affected by the oxidation state of the metal. For chromium, the differential curve J Cr+3 J Cr+6 (Fig. ) has a simpler structure. As the spectrum of chromate lacks the Cr Kβ line, the long-wave end of the curve reveals only one peak corresponding to that line of chromium oxide. Approximating the line shape with a Gaussian curve and assuming the shift to be small compared to the line width H, the shift may be approximately calculated as = 0.35H(A max А min )/A), where (A max А min )/A is the ratio of the difference between the maximum and the minimum of the S-shaped differential curve to the magnitude of the main peak. The values E, H and in this and following equations can be expressed in units of wave-length (Å) or energy units (ev), what is more convenient for estimation of a chemical shift. The use of this formula for the examples shown in Figs. 1 and yields the shift of 0.17 ev between Fe O 3 and FeC O 4, and 0.3 ev between K CrO 4 and Cr O 3. The largest shift was observed for uranium, about 1.3 ev. Similar results were obtained by evaluating the shift from the count rate ratio on the line slopes. Using the Gaussian approximation of the line shape, we can write N 1 E k = = EXP[ ], or E = ln k (1) N 0 where N is the count on a line slope at the distance Е (in energy units) from the peak and N 0 is the maximum peak count, is the dispersion which can be expressed as H H1/ = 1/ = () ln.355 Consider the ratio of pulse counts N and N on the right and left slopes of the reference line at Е and E from the peak, respectively. 0 = N = 1 E E EXP[ [ ]] N (3) If the line is shifted by (in the << Н limit), the new ratio may be expressed as N N 1 E+ E = = EXP[ [ ]] 1 (4) Then the logarithm of the ratio 1 / 0 is 1 ln = ln = ( E + E ) = ( E 0 Using (1), we can then write + E ) (5)

6 Copyright (c)jcpds-international Centre for Diffraction Data 00, Advances in X-ray Analysis, Volume ISSN = ln ( H 1/ ln ln k + ln k ) (6) or, assuming k and k to be approximately equal, = 8 H 1/ ln ln *ln k (7) A shift measurement error E is caused by the error of, which accordingly depends on the measurement errors of count rates on the slopes of the line. The latter errors consist of several components: statistical variation of pulse count, goniometer position reproducibility, sample reproducibility, and instrumental error of measurement. The contribution of each factor depends on the spectrometer characteristics and experimental strategy. In the present study of line shift using SPAK-1M spectrometer for multiple alternate measurements of each sample with repeated repositioning of the goniometer to the line slopes, the main factors of error were statistical variation of the count and goniometer position error. Taking a derivative of (7) over and assuming the statistical error of to be ( }stat = (8) k N 0 (as the four values of N are very close), and the goniometer positioning error to be 1 dn 4 ln *ln k ( )pos = * * E = E (9) k N de H 0 1/ K where K is the number of repositions, we can derive the respective error components for the shift: E E (stat) = (10) ln k N 0 and 8 E ln *ln k E (pos) = E (11) H1/ K Substituting (7) into (10) and (11), one can see that the statistical component is proportional to the line width, and the goniometer positioning component does not depend on instrument resolution. Table 1 shows the line shifts measured for the same sets of samples and the estimates for the component content assay error, С, calculated as c = %

7 Copyright (c)jcpds-international Centre for Diffraction Data 00, Advances in X-ray Analysis, Volume ISSN Table 1 Estimated errors of the oxidation state assay from the count rate ratio on two wavelengths. Compounds Wavelength, A atio, ev, ev С,,% λ λ = 1 / 0 Fe O 3 FeC O K CrO 4 Cr O UO 3 UO The table shows the error of oxidation state analysis using the SPAK-1M spectrometer to lie within 3 5%, which corresponds to the data from the analysis of the oxidation states of iron by measuring the shift of -series lines using the CAMEBAX micro-analyzer [5]. CONCUSION High goniometer positioning reproducibility of the SPAK-1M spectrometer allows using this instrument to study chemical shifts of X-ray lines. A possibility of oxidation state study by comparison of the K-spectra of transition metals and uranium -spectrum has been demonstrated. When different oxidation states of a transition metal are present in a sample, their content can be estimated within 3 5% accuracy from the ratio of count rates on the slopes of the Kβ line. EFEENCES 1. Brytov I.A., Obolensky E.A., Goldenberg M.S., abinovich.g., Antoyeva T.M., Magdin Yu.A. Apparatura i metody rentgenovskogo analiza (Equipment and methods of X-ray analysis), eningrad, 1983, No. 9,p Blokhin M.A., Nikiforov I.Ya. Apparatura i metody rentgenovskogo analiza (Equipment and methods of X-ray analysis),. eningrad, 197, No 10,p i Z.; uqin Y.; Shi.; Wang Q. Anal. Chim. Acta, 15 Jul 1991, 48 (1), Yoichi Tamaki X-ray Spectrom.,,1995,4, Hoffer H.E., Brey G.P., Schulz-Dobrick B., Oberhansli. Eur J. Mineral., 1994, 6,