Neutron Activation Analysis of Ultrahigh-Purity Ti Al Alloys in Comparison with Glow-Discharge Mass Spectrometry

Transcription

1 Materials Transactions, Vol. 43, No. 2 (2002) pp. 116 to 120 Special Issue on Ultra-High Purity Metals (II) c 2002 The Japan Institute of Metals Neutron Activation Analysis of Ultrahigh-Purity Ti Al Alloys in Comparison with Glow-Discharge Mass Spectrometry Atsushi Kimura 1, Seiichi Takaki 2, Yukihiro Naka 3, Yoshihiko Hayashi 3, Yuji Hori 1 and Kenji Abiko 2 1 National Institute of Advanced Industrial Science and Techlogy, Ikeda , Japan 2 Institute for Materials Research, Tohoku University, Sendai , Japan 3 Research Reactor Institute, Kyoto University, Sennan-gun, Osaka , Japan Neutron activation analysis (NAA) of ultrahigh-purity Ti Al alloys has been performed without chemical treatments using rmal and graphite-shielded irradiations. The Ti Al alloys were made of ultrahigh-purity Ti ( mass% or mass%) and Al ( mass%) by melting in ultrahigh vacuum with a copper-crucible induction-heating furnace. Concentrations and detection limits were evaluated for 35 impurity elements analyzed also by glow-discharge mass spectrometry. Seven impurity elements As, Cu, Ga, Na, Mn, Sb and W were detected by NAA. The sensitivities of NAA for Ga, In, Mn, Sb and Sc were better than those of glow-discharge mass spectrometry. Differences between the two analytical methods were mostly acceptable. The effectiveness of the graphite-shielded irradiation to suppress fast neutron reactions was clearly demonstrated. (Received November 2, 2001; Accepted December 21, 2001) Keywords: neutron activation analysis, titanium aluminum alloy, ultrahigh purity, glow discharge mass spectrometry, metallic impurity, detection limit 1. Introduction Recent studies on ultrahigh purification of base metals have revealed that some of their properties considered so far to be intrinsic are impurity effects. Ultrahigh-purity metals sometimes exhibit unkwn intrinsic properties which have the possibility of new industrial applications. For the developments of purification of metals, it is essential to identify trace amount of impurity elements and determine their concentrations with sufficient accuracy. Glow-discharge mass spectrometry (GDMS) is often used for the analysis of trace amount of metallic impurities in metals. However, GDMS requires standard samples having the same components as samples for accurate analysis, because relative sensitivity factors vary depending on matrix elements. The standard samples are t always available. Neutron activation analysis (NAA) is, in principle, one of absolute determination methods; that is, NAA requires standard samples once irradiation and nuclear parameters are precisely determined. Therefore, NAA can be used to verify analytical results obtained by GDMS. On the other hand, in the case of NAA of metals, gamma-rays originating from matrices often interfere with impurity signals. Chemical separation of interfering radioisotopes is, thus, required to achieve the best sensitivity. Radiochemical neutron activation analysis (NAA with chemical separation) has been applied for Ti 1, 2) and Al 3, 4) and sensitivities of ppb levels were obtained. However, the complicated chemical separation processes may cause errors in the same way as chemical analysis. Therefore, the present study examines the neutron activation analysis without chemical separation (i.e., instrumental neutron activation analysis) for ultrahigh-purity Ti Al alloys with different purity levels. Recent efforts on purification of Ti Al alloys succeeded in reducing gaseous impurities (C, N, O and S) from several hundreds mass ppm to several tens mass ppm by melting under ultrahigh vacuum conditions. 5 7) It is interesting to analyze other impurities such as metallic and metalloide impurities in the ultrahigh-purity Ti Al alloys. By analyzing 35 impurity elements in the Ti Al alloys, we compare NAA and GDMS to examine the validity of analytical results of the two methods. 2. Experimental Procedure Two grades of high-purity Ti ( mass% and mass%) and high-purity Al ( mass%) were used as starting materials. Note that the concentration of gaseous impurities were excluded from the above expression of purity. High purity Ti Al ingots of about 0.8 and 1 kg were prepared by melting the high-purity Ti and Al twice or 6, 7) three times in a copper-crucible induction-heating furnace. The starting materials were first melted in Ar gas atmosphere to alloy them homogeneously. The ingots were then melted in ultrahigh vacuum to remove the gaseous impurities. The pressure before melting was Pa and the pressure during the melting varied between Pa and Pa. As a result of melting, three sets of ingots were obtained. Ingot-1 was made of mass% Ti and Al. Ingot-2 was made of mass% Ti and Al but it was contaminated during the melting process. Ingot-3 was made of mass% Ti and Al. The final compositions were Ti 47 mol%al for Ingot-1/Ingot-2 and Ti 45 mol%al for Ingot-3. Thirty-five impurity elements in the samples were analyzed by GDMS using a VG-Microtrace VG9000 instrument as shown in Table 1. Since chemical analysis of metallic impurities in Ti Al is very difficult, it is t easy at present to obtain (high-purity) Ti Al standard samples analyzed accurately. No calibration standard was thus used for the present analysis and the results were corrected by relative sensitivity factors provided for the instrument except for Pt. The concentrations of Pt were estimated from ion current intensities.

2 Neutron Activation Analysis of Ultrahigh-Purity Ti Al Alloys in Comparison with Glow-Discharge Mass Spectrometry 117 Table 1 GDMS analysis of ultrahigh-purity Ti Al (mass ppm). Element Ingot-1 Ingot-2 Ingot-3 (4N5 Ti + 5N8 Al) (6N Ti + 5N8 Al, (6N Ti + 5N8 Al) Contaminated) Ag < 0.01 <0.01 As Not detectable Not detectable Not detectable B < 0.01 <0.01 Ba < 0.01 <0.01 Be < 0.01 <0.01 Bi < 0.01 <0.01 Ca < 1 <1 Cd < 0.1 <0.1 Co Cr Cu Fe Ga < <0.05 Ge < 0.05 <0.05 Hf < 0.05 <0.05 Hg < 0.05 <0.05 In < K < Li < 0.01 <0.01 Mg < <0.05 Mn <0.01 Mo < 0.1 <0.1 Na < Ni P 0.02 <0.01 Pb < 0.01 <0.01 Pd < 0.01 <0.01 Pt < 0.01 <0.01 Sb 4.6 <0.05 Sc < 0.2 <0.2 Si Sn 0.52 <0.05 V W < Zn < 0.1 <0.1 Total <21.51 <8.67 Arsenic is t detectable owing to the interference of matrix signals. The ingots were cut into small pieces ( g) of sample for neutron activation analysis. These samples were electrochemically polished in a solution of acetic acid (CH 3 CHOOH) and 7% perchloric acid (HClO 3 ) with a bias voltage of 58 V for 290 s, followed by rinsing in pure water and ethal. The samples were then separately packed in clean polyethylene bags and irradiated with neutrons. Three samples per ingot were analyzed each time (n = 3) and the same analysis was repeated twice for most of the samples. After neutron irradiation, the samples were taken out from the bags and separately put into new polyethylene bags to avoid gamma-rays from the irradiated bags. Neutron irradiation was performed at the Kyoto University Reactor (KUR) with two types of irradiation facilities. The pneumatic tube Pn-2 was used for rmal irradiation with thermal neutron flux of n/cm 2 and fast neutron flux Fig. 1 Gamma-ray spectrum for the sample from Ingot-1 after the rmal irradiation. Solid lines indicate real impurity peaks. Dotted lines indicate peaks originating from the matrix Ti Al and a background peak ( 40 K). of n/cm 2. The thermal column pneumatic tube TC- Pn was used for graphite-shielded irradiation to suppress fast neutrons. Thermal neutron flux and fast neutron flux were and n/cm 2, respectively. The Ti Al samples and pure iron (0.01 g) as a flux monitor were put into a polyethylene capsule designed for these irradiation tubes and transferred between an end station and a reactor core by using pressurized CO 2 gas. Gamma-rays were measured by a pure Ge detector surrounded by lead blocks. Energies and efficiencies of the detector were calibrated by using 11 standard radioactive isotopes. 3. Results and Discussion 3.1 Normal Irradiation Half of the Ti Al samples were first neutron-irradiated for 30 min by the rmal irradiation facilities and gamma-rays were measured after a cooling period of h. Figure 1 shows the typical gamma-ray spectrum of the sample from the Ingot-1. Gamma-rays were measured for 300 s. Many peaks were observed, but most of them arose from matrix Ti Al itself. Radioactive isotopes 46 Sc, 47 Sc and 48 Sc are produced from 46 Ti, 47 Ti and 48 Ti, respectively, in the matrix by fast neutron reactions. Radioactive 24 Na are produced from 27 Al in the matrix. Two peaks at and kev are escape peaks (EP) of the kev peak of 24 Na. Environmental (background) 40 K was detected at kev. Such gammarays clearly increased the background level of the spectrum through the Compton scattering. Only 76 As and 122 Sb peaks were detected as impurity peaks. Table 2 shows the concentrations of detected impurities by NAA analysis in comparison with the corresponding GDMS analysis (shown in Table 1). Two sets of samples (n = 3) taken from almost the same positions of the ingot were analyzed and the average values of the three samples for each irradiation are shown. We te that NAA can detect As which cant be detected by GDMS. The Ga value detected by NAA in Ingot-2 was similar to that by GDMS. The Sb values showed a difference between the 1 st and 2 nd runs and also a difference between NAA and GDMS. The 122 Sb peak

3 118 A. Kimura et al. Table 2 NAA analysis of ultrahigh-purity Ti Al by rmal irradiation (mass ppm). Element NAA GDMS 1 st run 2 nd run Ingot-1 (4N5 Ti + 5N8 Al) As Not detectable Sb Ingot-2 (6N Ti + 5N8 Al, contaminated) Ga Sb Ingot-3 (6N Ti + 5N8 Al) Sb <0.05 Table 3 NAA analysis of ultrahigh-purity Ti Al by graphite-shielded irradiation (mass ppm). Element NAA GDMS 1 st run 2 nd run Ingot-1 (4N5 Ti + 5N8 Al) As Not detectable Cu Ga <0.05 Na <0.01 Mn Sb W <0.01 Ingot-2 (6N Ti + 5N8 Al, contaminated) Cu 0.24 Ga Na Mn Sb W Ingot-3 (6N Ti + 5N8 Al) Cu Na Mn <0.01 Sb - - <0.05 W of 1000 mass ppm. Matrix-origin 24 Na and 46 Sc produced by fast neutron reactions strongly interfere with impurity signals. It is thus impossible to measure impurity concentrations of Na and Sc in this case. No impurity peak was found in the gamma-ray spectrum for the samples of Ingot-3 and hence there is open triangle in Fig. 2. On the other hand, most detection limits of NAA are worse than those of GDMS, while the detection limits of NAA for As, Ga, Mn, Sb and W are below 1 mass ppm close to those of GDMS. Fig. 2 Detection limits of NAA by the rmal irradiation for 35 elements in Table 1. Closed and open triangles are detection limits and detected values of NAA, respectively. Closed and open circles are detection limits and detected values of GDMS, respectively. (564.2 kev) appears very close to the 75 As peak (559.1 and kev). One possibility is that accurate peak separation was difficult due to statistical errors. The two peaks used for the Sb and As analysis are most intense among gamma-ray peaks corresponding to activated Sb and As, but it might be possible to choose other weaker peaks and to improve the accuracy of the analysis. In Fig. 2, detection limits of 35 elements listed in Table 1 were calculated for Ingot-3 (see Appendix A), including the impurity elements which were t detected by NAA. Data points of the NAA detection limits were connected by broken lines for a plain sight. When there is appropriate neutron reactions for NAA or a calculated detection limit is more than 1000 mass ppm, the data point is plotted on the line 3.2 Graphite-Shielded Irradiation A problem with the rmal irradiation is fast neutron reactions which induce unnecessary radioisotopes from matrix elements (i.e., Ti and Al). Such radioisotopes produce strong gamma-rays which increase background signals and saturate detector electronics. To solve this problem, we used the graphite-shielded pneumatic tube with a low fast neutron flux. Indeed, the effectiveness of the graphite-shielded irradiation was successfully demonstrated for ultrahigh-purity Fe. 8) By using the graphite-shielded pneumatic tube, the ratio (thermal neutron flux)/(fast neutron flux) can be improved from 4.7 to 5000, by three orders of magnitude. The thermal neutron flux also decreased approximately by two orders of magnitude due to the graphite shield. Therefore, the total irradiation period was increased from 30 min to 24 h to get a similar thermal neutron dose. Figure 3 shows the gamma-ray spectrum of the sample from Ingot-1. The gamma-rays were measured for 500 s per one sample. It is clear that most of the Sc peaks originat-

4 Neutron Activation Analysis of Ultrahigh-Purity Ti Al Alloys in Comparison with Glow-Discharge Mass Spectrometry 119 Fig. 3 Gamma-ray spectrum for the sample from Ingot-1 after the graphite-shielded irradiation. Solid lines indicate real impurity peaks. Dotted lines indicate a peak originating from the matrix Ti and a background peak ( 40 K). ing from the matrix Ti disappeared, and that the background level decreased by one order of magnitude around 1000 kev and 2500 kev. Peak heights of Na decreased and hence the Na escape peaks observed in Fig. 1 also disappeared. New peaks corresponding to Ga, Mn, Cu appeared. They were most likely below the background level in the case of the rmal irradiation. The concentrations of detected impurities were summarized in Table 3 and compared with the corresponding GDMS results shown in Table 1. The NAA results in the first and the second runs were consistent with each other. The differences between NAA and GDMS are mostly acceptable. In the case of the rmal irradiation, the concentration of Na was calculated to be around 80 mass ppm on the assumption that all of 24 Na isotopes were produced by the thermal neutron reactions. The ratio of the thermal neutron flux to the fast neutron flux fell to 0.1% for the graphite-shielded irradiation. As a rough approximation, 24 Na radioisotopes produced by the fast neutron reaction also decreased to 0.1% (i.e., 0.08 mass ppm). If 0.08 mass ppm is subtracted from the Na concentration measured by NAA, then the obtained concentrations appear to be more reasonable. The suppression of fast neutrons in the present experiment is t yet eugh for this low level of concentration; however, the graphiteshielded irradiation made it possible to roughly estimate the Na concentration. Detection limits for the samples of Ingot-3 measured after the graphite-shielded irradiation were calculated as shown in Fig. 4. Compared with Fig. 2 for the rmal irradiation, the detection limits were substantially improved for most of elements. Sc radioisotopes produced from Ti by the fast neutron reaction were negligible in the graphite-shielded irradiation. A 47 Sc peak is still observed at kev, but there is 46 Sc peak (889.3 kev) which is used to calculate impurity Sc concentrations. It means that the concentration of impurity Sc can be analyzed without the interference of matrix Sc signals. In the case of Ga, In, Mn, Sb and Sc, the detection limits of NAA are better than those of GDMS. In particular, the detection limit of In was significantly improved by more than five orders of magnitude. This is mainly because Fig. 4 Detection limits of NAA by the graphite-shielded irradiation for 35 elements in Table 1. Closed and open triangles are detection limits and detected values of NAA, respectively. Closed and open circles are detection limits and detected values of GDMS, respectively. of the difference in cooling period. The cooling periods for the rmal and graphite-shielded irradiations were h and 2 3 h, respectively. Indium radioisotope ( 116m1 In) used for the calculation has a half life of m. The difference in cooling period of 15 h gives rise to the decay of gammarays approximately by five orders of magnitude. The use of the graphite-shielded irradiation is effective t only to suppress the fast neutron reaction but also to optimize the irradiation condition (i.e., combination of irradiation and cooling periods). The present study aims to examine the ability of instrumental NAA (NAA without chemical separation) and to verify results of GDMS often used in metal purification studies. The results satisfied this purpose for some of impurity elements as shown in Table 3. The sensitivity of GDMS depends on both matrix elements and compositions, while that of NAA depends on neither matrix elements r compositions. GDMS requires calibration standards every time for different types of matrix. Hence, NAA has a big advantage over GDMS. In the present study, however, the impurity concentration was calculated based on the nuclear parameters in published data tables. Sometimes irradiation and nuclear parameters used for NAA are t sufficiently reliable for quantitative analysis. For more precise measurements, it is necessary to do calibration only once for each impurity element with an appropriate calibration method. 9) 4. Conclusion Neutron activation analysis without chemical treatments was applied for ultrahigh purity Ti Al alloys and compared with analytical results obtained from glow-discharge mass spectrometry for 35 impurity elements.

5 120 A. Kimura et al. (1) Seven impurity elements As, Cu, Ga, Na, Mn, Sb and W were detected. (2) Sensitivities for Ga, In, Mn, Sb and Sc were better than those of GDMS. (3) Impurity As which cant be detected by GDMS was successfully analyzed. Compared with GDMS, NAA showed acceptable results for both concentrations and detection limits of impurities. The effectiveness of graphite-shielded irradiation to suppress fast neutron reactions was clearly demonstrated. Ackwledgements The authors gratefully ackwledge the support from the Ministry of Education, Culture and Science and the Program of Core Research Evolution Science and Techlogy (CREST), Japan Science and Techlogy Cooperation for the present study. We wish to thank Mr. T. Nakajima and Mr. C. Kawarada (IMR) for Ti Al sample preparation, and Mr. N. Tsubouchi (AIST) and Prof. T. Yoshiie (KURRI) for assistance on neutron activation analysis. We also thank Nikko Materials Co., Ltd. for GDMS analysis. REFERENCES 1) K. S. Park, N. B. Kim, K. Y. Lee, Y. Y. Yoon, S. K. Chun and J. H. Lee: J. Radioanalytical and Nucl. Chem. 192 (1995) ) D. Wildhagen and V. Krivan: Anal. Chem. 67 (1995) ) T. Mitsugashira, Y. Koma, S. Hirai, I. Okada, N. Kurashima and H. Sakurai: J. Radioanalytical and Nucl. Chem. 147 (1991) ) J. H. Zaidi, M. Arif. I. Fatima, S. Ahmed and I. H. Qureshi: J. Radioanalytical and Nucl. Chem. 241 (1999) ) T. Nakajima, Y. Morimoto, S. Takaki and K. Abiko: Phys. Status. Solidi. (a) 167 (1998) ) T. Nakajima, Y. Morimoto, S. Takaki and K. Abiko: Mater. Trans., JIM 41 (2000) ) C. Kawarada, N. Harima, S. Takaki and K. Abiko: Phys. Status. Solidi. (a) in press. 8) A. Kimura, S. Takaki, Y. Naka, Y. Hayashi, Y. Hori and K. Abiko: Mater. Trans., JIM 41 (2000) ) A. Simonits, L. Moens, F. De Corte, A. De Wispelaere, A. Elek and J. Hoste: J. Radioanalytical Chem. 60 (1980) Appendix Detection limits were calculated as follows. We assumed that (1) impurity signal must be more than twice the statistical error of a background signal and (2) a width of gamma-ray peaks is 5 kev. The background signal in the measured spectrum was integrated in the rage of 5 kev at the peak energy of a target isotope. When the integrated count is N B, the statistical error σ B can be expressed as σ B = N 1/2 B. From the assumption (1), 2σ B is a minimum detectable signal. A con- Table A I Isotopes used for the calculation of detection limits. Element Isotope Peak Half life Note energy (kev) Ag Ag-110m d As As h B Ba Ba-135m h Be Bi Bi-210m E+06 y Ca Ca d Cd Cd h Co Co y Cr Cr d Cu Cu h Fe Fe d Ga Ga h Ge Ge h Hf Hf-180m h Hg Hg d In In-116m m K K h Li Mg Mg m Mn Mn h Mo Mo h Na Na h Ni Ni h P Pb Pd Pd h Pt Pt d 1 Pt m 2 Sb Sb d Sc Sc d Si Si h Sn Sn d V V m W W h Zr Zr h (y: year, d: day, h: hour, m: minute) 1 Normal irradiation 2 Graphite-shielded irradiation centration value can be calculated from this 2σ B with other irradiation parameters and data on a calibration standard. When more than one radioisotope are produced from one element, the radioisotope with the best detection sensitivity was selected. Table A I summarizes the radioisotopes used in the present study together with peak energies and half lives.