Investigation of impurities in thermoluminescent Al2O3 materials by prompt-gamma activation analysis† Zs. Kasztovszky,*a Zs. Re�vay,a T. Belgya,a B. Fazekas,a J. O� sto�r,a G. L. Molna�r,a G. Molna�rb and J. Borossayb aDepartment of Nuclear Research, Institute of Isotope and Surface Chemistry, Chemical Research Center, POB 77, H-1525 Budapest, Hungary. E-mail: kzsolt@alpha0.iki.kfki.hu bDepartment of General and Inorganic Chemistry, Eo�tvo�s Lora� nd University, POB 32, H-1518 Budapest 112, Hungary Received 11th December 1998, Accepted 21st January 1999 a-Al2O3 is one of the most important materials for thermoluminescence dosimetry.The thermoluminescent features are strongly aVected by impurities, which were investigated with the non-destructive method of prompt-gamma activation analysis (PGAA). Impurities in four samples of thermoluminescent alumina materials were investigated. The precision and reproducibility of the PGAA method were also tested on standard samples spiked with B, Na, S, Cl, Fe, Cu and Ag.low thermal eVective flux (2.5×106 cm-2 s-1), destruction Introduction eVects on the crystal lattice are negligible; thus, the PGAA High-purity aluminium oxide powders are widely used for the method preserves the original physical and chemical form even production of diVerent ceramics. Impurities, even in trace of single crystal samples. Moreover, one avoids any sample amounts, can aVect the quality (i.e. optical, electrical, mechan- preparation problems arising from the chemical and mechanical ical and other properties) of these ceramics.Recently, specially resistance of aluminium oxide. prepared alumina ceramics (and single crystals also) have been Because only mass ratios of the elements in a sample are increasingly applied in the field of thermoluminescence (TL) determined, matrix eVects due to self-absorption and scattering dosimetry. Since impurities can be directly involved in the TL of neutrons are cancelled.10 Moreover, aluminium oxide is an mechanism, they can determine basic dosimetric properties ideal matrix for PGAA because aluminium and oxygen have such as sensitivity and fading.Therefore, the determination of medium and low neutron capture cross-sections, respectively, impurities in these materials is important. so trace elements are well distinguished on the low spectral Chemical analysis of Al2O3 powders can be performed by background. This is in contrast to INAA where a high level direct or solution-based methods.Among the direct methods, of short-lived activity is produced due to the high thermal flux optical emission spectrometry is the most widespread.1 This and the presence of fast neutrons.7 method, especially when alkali halides are employed as thermo- The aim of this work was to demonstrate the performance chemical reagents, is sensitive to many important elements, of the PGAA method for the analysis of thermoluminescent but suVers from poor reproducibility and requires solid refer- alumina materials.These measurements—among other ence materials for calibration. Other methods such as slurry parallel analytical experiments—serve as test cases for the sampling, direct sample insertion and electrothermal vapori- applicability of PGAA with guided neutrons. zation techniques, combined with ICP-AES/MS, generally oVer better detection limits, but they have several limitations Experimental (e.g.calibration problems, grain size limitations).2,3 When solution-based methods, such as ICP-AES/MS and AAS, are PGAA method applied, a complete dissolution of the solid sample is manda- Prompt-gamma activation analysis is based on the detection tory.4,5 This results in higher detection limits and lower of prompt gamma rays originating from neutron radiative precision because of the dilution and contamination of the capture or (n, c) reaction. By the spectral analysis of prompt sample.6 Moreover, owing to the high melting-point as well gamma rays, according to their energies and intensities, it is as the hardness and chemical inertness of alumina, the dissolupossible to determine the chemical composition of a given tion is tedious and time-consuming. sample.The nuclear reaction is independent of the physical Although instrumental neutron activation analysis (INAA) is and chemical form of the samples. a well-established technique and has been widely applied to the analysis of alumina,7 prompt-gamma activation analysis (PGAA) is a relatively new technique, which can be performed Experimental apparatus.The measurements were carried out best with cold neutron beams of nuclear reactors.8,9 PGAA at the Budapest Research Reactor, Hungary. This is a waterenables, in principle, the detection of all elements of the Periodic cooled, water-moderated research reactor with a thermal power Table — many of them with remarkable sensitivity.Since the of 10MW. A beam of low energy neutrons is transported to a neutrons used for PGAA are of low energy (<25 meV) and of distance of approximately 35 m from the reactor core by a slightly curved neutron guide, made of glass coated with a nickel reflector. At the end of the guide—where the prompt-gamma †Presented at the 8th Solid Sampling Spectrometry Colloquium, Budapest, Hungary, September 1–4, 1998. experimental apparatus is situated—the energy of the neutrons J.Anal. At. Spectrom., 1999, 14, 593–596 593is less than that of thermal neutrons, because the higher energy the quality factor (Q), which is derived from the diVerence between the measured and literature energies. The closer the neutrons are not reflected. The thermal-equivalent eVective flux is approximately 2.5×106 cm-2 s-1 at the target position. The value of Q is to unity, the more reliable is the element identification. One must also consider the intensity ratios system has been described elsewhere.11,12 The samples are usually sealed in thin FEP Teflon bags or between the diVerent gamma peaks of an element. The possible background lines are also examined.The background gamma Teflon capsules, and placed directly in the beam, which is collimated to an area of about 2×2 cm. Because the sample is rays mostly originate from the (n, c) reactions in the surrounding material. The most important sources are H and N in the virtually transparent to neutrons, the determined chemical composition is an average value for the whole irradiated volume.air, Fe and Al in the equipment’s material and F in the Teflon packaging of the samples. Natural background radiation can The emitted prompt gamma rays are observed with a complex detector system. The main part of this system is a also be observed in the spectra from members of the U and Th series. Canberra high-purity germanium detector. The detector is surrounded by a bismuth germanate (BGO) scintillator annulus in order to reject the Compton-scattered photons.With Determination of chemical composition. The detected gammaray intensity, represented by the peak area (AE), is directly the present set-up, a baseline suppression ratio of about six can be achieved for the 1332 keV energy peak of 60Co. This proportional to the mass of a given chemical element, and the measuring time, t. Hence ratio becomes much higher for higher energy c-rays.9 The whole detector system is shielded against neutron- and gammaradiation background.m= 1 S · AE t (1) The spectra are collected into 16 000 channels by a Canberra S100 MCA multichannel analyzer. The energy and eYciency where calibrations of the measuring system are performed using precisely known c-lines emitted by radioactive sources and S= NA M ·h·s0·Ic·W0·e(Ec) (2) (n, c) reactions. The collected spectra are evaluated by ‘Hypermet PC’, a complex c-spectrum evaluation program, is the analytical sensitivity, expressed in units of developed in our laboratory.13 counts s-1 mg-1.It is proportional to the neutron capture cross-section of the nucleus s0, the isotopic abundance h and Element identification. The element identification is based the gamma yield Ic, which are nuclear constants, as well as to on our own nuclear data ler compilation of the neutron flux W0 and the detector eYciency e(Ec), which (n, c) data, compiled by Lone et al.,14 is being replaced by new are characteristics of the measuring system. The sensitivity for experimental data measured with higher precision in our an element can be related to that of a comparator element: laboratory for each stable element.The elements are identified according to the energy values of their most intense prompt- Sx SC = (h·s0·Ic/M)x (h·s0·Ic/M)C · ex(Ec) eC(Ec) = k0,C(x)· ex(Ec) eC(Ec) (3) gamma peaks. The reliability of the element identification is controlled through diVerent statistical parameters, which are This ratio is independent of the neutron flux; it depends only derived from the deviations of measured energy and intensity on nuclear constants and the detector eYciency.The latter are values from the literature values.15 One of these parameters is known with good accuracy. The k0-factors were determined by internal standardization measurements. The mass ratio for Table 1 Interference-free prompt-gamma lines used for determination an element ‘x’ can be determined according to the following of mass fractions.k0-factors are related to the 1950.955 keV promptequation: gamma line of Cl Element Energy/keV k0 Uncertainty of k0 (%) mx mR = AE,x AE,R · k0,C(R) k0,C(x) · eR(Ec) ex(Ec) (4) H 2223.182 1.801 0.65 where R is an arbitrary reference element contained in the B 478.000 3.597×10-2 0.93 sample. The comparator elements were typically H and Cl in Na 869.160 1.424×10-2 1.1 samples with precisely known compositions.10,16 874.369 9.946×10-3 1.28 Al 1778.888 8.297×10-2 2.27 The masses were calculated according to eqn.(1)–(4). The 2271.617 1.414×10-3 4.14 gamma lines used and the corresponding k0-factors are listed 2282.705 3.179×10-3 2.96 7723.645 2.286×10-2 2.5 Table 2 Sensitivities and detection limits for the present PGAA system S 840.928 3.212×10-2 1.57 3220.468 1.113×10-2 2.59 4869.297 5.632×10-3 3.28 Sensitivity/ Detection limit/ Element counts s-1 mg-1 mg g-1 Cl 1164.740 1.38 0.86 1950.955 1 1959.146 6.428×10-1 1.46 H 6.49×10-2 13 B 4.02×101 0.8 2863.750 2.899×10-1 0.92 6110.692 1.063 1.25 N 2.37×10-4 6100 F 1.38×10-4 17 600 Fe 352.326 2.641×10-2 2.44 691.876 1.357×10-2 2.51 Na 1.78×10-2 570 Al 2.22×10-3 520 1612.682 1.542×10-2 2.83 5920.323 2.340×10-2 3.06 S 4.28×10-3 290 Cl 8.10×10-2 15 7631.123 6.903×10-2 2.83 7645.580 5.805×10-2 2.90 Ca 1.75×10-3 520 Fe 3.84×10-3 315 Cu 203.000 1.568×10-2 1.68 278.273 7.222×10-2 1.33 Cu 1.18×10-2 150 Zn 1.88×10-3 680 7306.598 1.970×10-2 2.29 7915.419 7.013×10-2 2.17 Ga 1.73×10-3 1070 Ag 6.86×10-2 29 Ag 237.100 8.774×10-2 2.08 360.632 7.013×10-2 2.97 Eu 1.16×101 0.18 Gd 1.35×101 0.15 657.911 1.004×10-1 1.83 594 J.Anal. At. Spectrom., 1999, 14, 593–596Table 3 Mass fractions of elements in four Al2O3 samples measured by PGAA. The values and the 1s uncertainty are given in mg g-1 TLD-800 CA-600 Pure 0223 CA-320 Sample c±uncertainty Q c±uncertainty Q c±uncertainty Q c±uncertainty Q H <13 — 125±5.5 0.80 <13 — 300±9 0.97 B 1.76±0.05 0.46 <0.8 — 1.1±0.4 0.93 <0.8 — Na <500 — 1530±72 0.80 <570 — 4380±80 0.93 Cl 52±3.1 0.75 <15 — <15 — <15 — Fe <300 — 525±79 0.73 <315 — 590±94 0.65 Cu <150 — <150 — <150 — <150 — Zn <680 — <680 — <680 — <680 — Ga <1070 — <1070 — <1070 — <1070 — Ag <30 0.47 <30 — <30 — 220±33 0.62 Eu <0.1 — <0.1 — 0.1±0.06 0.82 <0.1 — Table 4 Mass fraction (%) and 1s uncertainties of spiked impurities measured by PGAA PGAA PGAA PGAA PGAA Element Compound Nominal measurement 1 measurement 2 measurement 3 average B H3BO3 0.034 0.0328±0.0006 0.0339±0.0006 0.0353±0.0006 0.0340±0.0006 Na NaCl 0.617 0.55±0.08 0.64±0.05 0.68±0.03 0.62±0.08 S Fe2 (SO4)3 0.519 0.54±0.02 0.59±0.02 0.57±0.02 0.57±0.02 Cl NaCl 1.026 0.98±0.02 1.01±0.02 1.07±0.02 1.02±0.02 Fe Fe2(SO4)3 0.490 0.49±0.02 0.58±0.02 0.52±0.02 0.53±0.02 Cu CuCl2 0.373 0.37±0.01 0.38±0.01 0.41±0.009 0.39±0.01 Ag AgNO3 0.326 0.330±0.009 0.324±0.009 0.336±0.009 0.330±0.009 in Table 1.The possible interferences between diVerent gamma and height 5 mm were pressed from the same homogeneous powder with a hydraulic press. These discs were sealed in peaks were individually examined, and the peaks aVected by spectral interference neglected. The elemental masses were Teflon bags in groups of three. The three diVerent samples with a mass of about 3 g each were measured individually for related to the mass of Al2O3, which was determined from the detected mass of Al by a calculation according to stoichi- 85 000–230 000 s.The analytical purity of the Al2O3 was checked by a PGAA measurement of a 4 g ‘pure’ Al2O3 ometry. Because of the poor detectability of oxygen, this procedure gives more precise mass values than the direct sample. Unexpectedly, it was found to be contaminated with Na (mass fraction 0.216%) and S (mass fraction 0.095%).measurement of the oxide. The total uncertainties (standard deviations) were calculated from the statistical (counting) During the calculations of nominal mass fractions, these values were taken into account. uncertainties of the peak areas, the uncertainties of the k0- factors and the uncertainties of the detector eYciencies. As the last two typically have standard deviations of a few per Results and discussion cent, the total uncertainty is mainly determined by the counting statistics reflected in the analytical sensitivities.Four alumina samples with previously unknown contaminants were investigated with PGAA. The mass fractions of the identified elements are shown in Table 3. The Q-factors show Detection limits. For the undetected elements, detection limits (CL) were calculated from the spectra themselves the reliability of element identification in PGAA. It is a remarkable feature of PGAA that the detection limit for some light elements, such as B, is much lower than that of ICP.CL= 3·sB S (5) Moreover, H cannot even be detected by ICP. For the samples with known contaminants, the nominal where sB is the standard deviation of the baseline and S is the mass fractions and the results of PGAA measurements for sensitivity for a given element. The baseline is a complex some elements are compared in Table 4. The elements investi- function, calculated by Hypermet PC automatically for each region.sB is the constant part of the function. Table 2 contains the sensitivity values and the calculated detection limits of the elements investigated. Measurements. The elemental compositions of three commercial alumina TLD powders, namely TLD-800 (VICTOREEN), CA-600 and CA-320 (DESMARQUEST, France) were investigated by means of PGAA. The masses of the individual samples were between 1 and 3 g. The samples were sealed in thin Teflon bags and irradiated for 17 000–63 000 s.In order to check the reliability of the PGAA results for some elements, a nominally pure aluminium oxide powder sample with a known amount of contaminants was prepared in our laboratory. It was spiked with B, Na, S, Fe, Cu and Ag in the form of NaCl, Fe2(SO4)3·9H2O, CuCl2·2H2O, 6 B. Zs. Varga, M. Ballo� k, G. Molna�r and A. Bartha, unpublished gated were B, Na, S, Cl, Fe, Cu and Ag. Three measurements work. were performed. It was found that the three measurements 7 H.Rausch, S. To� ro�k and A. Simonits, Isotopenpraxis, , show good reproducibility, and the agreement between nom- 229. inal and measured mass fractions is good (see Fig. 1). 8 R. M. Lindstrom and C. Yonezawa, in Prompt Gamma Neutron Activation Analysis, ed. Z. B. Alfassi and Ch. Chung, CRC Press, Boca Raton, FL, 1995, p. 93. Conclusion 9 G. L. Molna�r and R. M. Lindstrom, in Nuclear Methods in Mineralogy and Geology, ed. A. Ve� rtes, S. Nagy and K. Su� vegh, Prompt-gamma activation analysis without the use of stan- Plenum Press, New York, 1998, p. 145. dards is a valuable analytical tool. It should be emphasized 10 G. L. Molna�r, Zs. Re�vay, R. L. Paul and R. M. Lindstrom, that the method is non-destructive and no sample preparation J. Radioanal. Nucl. Chem., 1998, 234, 21. is necessary. Although at present the detection limits for many 11 G. L. Molna�r, T. Belgya, L. Dabolczi, B. Fazekas, A� . Veres, I. elements are higher than those characteristic of other well- Bikit, Z.Kis and J. O� sto� r, J. Radioanal. Nucl. Chem., 1997, 215, established techniques (e.g. ICP-AES, INAA), some elements, 111. 12 T. Belgya, Zs. Re�vay, B. Fazekas, I. He�jja, L. Dabolczi, G. L. viz., H, B, Cl, Gd and Eu, can be detected at very low levels, Molna�r, Z. Kis, J. O� sto� r and Gy. Kasza�s, in Proceedings of the 9th which are diYcult to attain by other methods. International Symposium on Capture Gamma-ray Spectroscopy, ed. PGAA has been shown to be a useful complementary G.L. Molna�r, T. Belgya and Zs. Re�vay, Springer Hungarica, method for the determination of impurities in thermoluminesc- Budapest, 1997, p. 826. ent alumina materials. The analytical sensitivity should be 13 B. Fazekas, G. L. Molna�r, T. Belgya, L. Dabolczi and improved by the installation of a cold neutron source, as A. Simonits, J. Radioanal. Nucl. Chem., 1997, 215, 271. 14 M. A. Lone, R. A. Leavitt and D. A. Harrison, Atomic Data Nucl. expected from the experiences at the cold-neutron based PGAA Data Tables, 1981, 26, 511. facilities of NIST, USA,17 and JAERI, Japan.18 15 J. O� sto� r, Z. Kis, B. Fazekas, G. L. Molna�r and A. Simonits, in Proceedings of the 9th International Symposium on Capture References Gamma-ray Spectroscopy, ed. G. L. Molna�r, T. Belgya and Zs. Re� vay, Springer Hungarica, Budapest, 1997, p. 788. 1 W. Schro� n, M. Krieg, D. Wienke, M. Wagner and K. Danzer, 16 M. Heurtebise and J. A. Lubkowitz, J. Radioanal. Chem., 1976, Spectrochim. Acta, Part B, 1992, 47, 189. 31, 503. 2 Gy. Zaray, G. Konya, J. A. C. Broekaert and F. Leis, Chem. Anal. 17 R. L. Paul, R. M. Lindstrom and A. Heald, J. Radioanal. Nucl. (Warsaw), 1990, 35, 311. Chem., 1997, 215, 63. 3 Z. Slova�k and B. Docekal, Anal. Chim. Acta, 1981, 129, 263. 18 C. Yonezawa, Anal. Sci., 1993, 9, 185. 4 T. Ishizuka, Y. Uwamino, A. Tsuge and T. Kamiyanagi, Anal. Chim. Acta, 1984, 161, 285. 5 E. Tata�r, I. Varga and Gy. Za� ray, Mikrochim. Acta, 1993, 111, 45. Paper 8/0885