Measurement of Neutron Capture Cross Sections with Fe-Filtered Beam*

Journal of NUCLEAR SCIENCE and TECHNOLOGY, 15[9], pp. 637~644 (September 1978). 637 Measurement of Neutron Capture Cross Sections with Fe-Filtered Beam* Nobuhiro YAMAMURO, Takeshi DOI, Toshiharu MIYAGAWA, Research Laboratory for Nuclear Reactors, Tokyo Institute of Technology** Yoshiaki FUJITA, Katsuhei KOBAYASHI and Robert C. BLOCKt Research Reactor Institute, Kyoto University*** Received April 3, 1978 The neutron capture cross sections of 93Nb, 11515In, 127I, 165H0, 181Ta, 232Th and 238U were measured using the Fe-filtered beam. A 15-cm thick Fe filter was placed in the neutron beam produced by the KUR 46-MeV electron Linac and capture g-rays were detected by two C6F6 scintillation detectors located at an 11.7 m-flight path. The pulseheight weighting technique was used to determine the relative capture g-ray detection efficiency. The neutron flux was measured by the same detectors, whose detection efficiency for the 480-keV g-ray from the 10B(n, a1g) reaction was calibrated by the saturated resonance capture in Ag at 5.2-eV. Self-shielding and multiple scattering corrections were applied to the data. The results of 24-keV capture cross sections are 340, 770, 780, 1,280, 880, 520 and 520 mb for 93Nb, 115In, 127I, 165Ho, 181Ta, 232Th and 238U, respectively. Total errors are 5 to 8%, with an estimated systematic error of 4%. The discrepancy between the present results and other data measured recently is within 10%. KEYWORDS: neutron capture cross sections, niobium 93, indium 115, iodine 127, holmium 165, tantalum 181, thorium 232, uranium 238, Fe-filtered neutron beam, data, errors I. INTRODUCTION A "point" cross section measurement using a filter is an attractive method to determine cross sections accurately because a low background neutron flux can be obtained. If a suitable filter which has a neutron cross section minimum is placed into a pulsed white or a reactor beam, a monochromatic neutron beam is transmitted from the neutron source to the experimental area. In case of a pulsed white beam produced by a target-moderator assembly bombarded by high energy electrons, the monochromatic neutron beam which has a peak at 24.3 kev and a full width at half maximum of about 2 kev was obtained after passing through an iron plate of 30-cm thickness(1). The background level near the 27-keV peak in the iron cross section was below about 0.15% of the peak counting rate at the 24.3 kev resonance-potential interference minimum. When a filter composed of 23-cm of iron, 35-cm of aluminum and 5-cm of sulfur was inserted into a reactor beam, a similar monochromatic beam has also been obtained(2). Using these monochromatic neutron beams, a precise absolute 24-keV cross section can be measured, and relative energy dependent measurements can be normalized to this "point" cross section to provide a n accurate determination of the cross section in the kev region. With the aid of the Fe-filtered beam, * This study was financially supported by the Visiting Researches Program of Research Reactor Institute, Kyoto University. ** Oh-okayama, Meguro-ku, Tokyo. *** Kumatori-cho, Sennan-gun, Osaka. Visiting Professor from Rensselaer t Polytechnic Institute, U.S.A. 1

638 J. Nucl. Sci. Technol., the neutron capture cross sections of In, Ta and depleted U were measured at the RPI 1.25-m diam. liquid scintillation spectrometer In the present experiment, the 24-keV neutron capture cross sections of 93Nb, 11515In, I, 165Ho, 181Ta, 232Th and 238U 127have been measured with a 15-cm thick Fe-filtered beam and a different g-ray detection technique. The preliminary result of 24-keV cross sections for several nuclei was given in the previous report(5). II. EXPERIMENTAL METHOD 1. Apparatus The experimental arrangement is shown in Fig. 1. The neutrons were generated by a tantalum photoneutron target with a 5-cm thick polyethylene moderator irradiated with electrons from the KUR 46-MeV linear accelerator. The accelerator was typically operated with a 0.6- µsec electron pulse width and a 140-Hz repetition rate. Fig. 1 Experimental arrangement of 24-keV capture cross section measurement A 5-cm thick iron filter was placed at the entrance to the evacuated flight tube, about 2.5 m from the neutron source. Another 10 -cm thick iron filter was placed at about a 7-m flight distance. The Fe-filtered neutron beam was collimated to 46-mm diameter onto a sample, which was mounted on a 4-position automatic sample changer. The neutron flight path was 11.7 m. Two C6F6 liquid scintillation detectors of 10-cm diameter by 4-cm thick were located face to face and were shielded from g-ray background by a 5-cm or a 10-cm thick lead wall. The signals from the scintillation detectors were fed into a linear amplifier and the amplified signals were passed to a timing single channel discriminator whose bias was set at about 150 kev. The linear signals and the timing logic signals were recorded with 32- channels X 32-channels pulse height and timeof-flight mode in a Nuclear Data 4420 Analyzer. 2. Measurements By placing a thick 10B sample between two C6F6 detectors, the Fe-filtered beam spectrum near 24 kev was measured. In this measurement, the electron pulse width was 60 nsec and the channel width of the time analyzer was 30 nsec. As the neutron capture probability of thick 10B sample is nearly constant for neutrons in the 24-keV band, the spectrum shown in Fig. 2 illustrates directly the spectrum of the neutron flux near 24 kev. Since the counting rate of C6F6 detectors for the -rays emitted when neutrons were captured g by nuclei was not high, a 15-cm thick iron filter was used so that the energy band width of the 24-keV filtered neutron beam was broader than that shown in the Refs. (1) and (2), and the ratio of peak count to the background count is about 200. The resonance shape dips are caused by resonances in the manganese impurity in the iron filter. 2 --

Vol. 15, No. 9 (Sep. 1978) 639 The straight line in the figure indicates the background counts which are determined by connecting both wings of the peak count. Resonance shape dips near the peak are caused by the manganese impurity in the Fe filter. cept iodine which was PbI2 powder capsuled into an aluminum can. The thickness for each sample used in the present experiment is shown in Table 1. The abundance of 285U in the sample of depleted uranium was less than 400 ppm. When the data were acquired, ordinarily three samples and the reference sample were mounted and samples were cycled into the neutron beam one by one by the sample changer such that a cycle was repeated every 15 to 20 min. The signals for each sample were stored automatically into the corresponding position of memories in the analyzer. Although the beam intensity varied gradually in the course of an experimental run of 10 to 20 hr, the total neutron flux impinging on samples was proportional to the irradiation time of each sample within less than 1% error, because the samples were irradiated repeatedly over many cycles. This fact was well-ascertained by a 20-hour test run. Figure 3 shows the time-of-flight spectra for 24-keV neutron capture in PbI2 and 238U Fig. 2 Fe-filtered neutron beam time-of-flight spectrum The incident neutron flux was measured with a 10B sample because the 480-keV r-rays emitted from the 10B(n, a1g) reaction can be detected with the C6F6 detectors. Boron sample was made of boron powder packed into an aluminum capsule. The sample was chemically analyzed for boron and was isotopically analyzed for 10B. It was found to be 93.93 B of which 93.27 W/0 was 10B. About W/0 17% of the incident 24-keV neutrons were captured by this sample. The counting rate of the 480- kev g-ray was apt to vary with the fluctuation of the discrimination level of about 150 kev during a long time experimental run. Thus, in the first run, a reference sample (a 2.5-mm thick Ag plate) was mounted in addition to the "B sample, and a careful comparison was made between the 10B and reference samples. After this run, the count for the reference sample was used as the neutron monitor. Metallic samples were used as targets ex- The channel width of the time analyzer used is 0.25 µsec. The regions between the solid lines represent the interval over which the data are integrated. Fig. 3 Typical time-of-flight spectra for 24-keV neutron capture in samples of PbI2 and 238U 3

640 J. Nucl. Sci. Technol., samples. Figures 4(a) and (b) show the pulse height spectra for each sample of PbI2 and 238U, respectively. In Figs. 4(a) and (b), the solid dots indicate the background subtracted pulse height spectra. These g-ray spectra were multiplied by the weighting function of the C6F6 detector obtained previously(6) according to the pulse-height weighting technique(7). The weighted spectra thus obtained were shown by the open circles in the figures. counting rate integrated over the 24-keV neutron band is given by the equation ( 1) where F means the neutron flux incident upon a sample, Y the average capture probability of a sample and e the g-ray detector efficiency. As well known, it is important for the capture cross section measurement that e is essentially independent of the g-ray cascade mode. With the use of the pulse-height weighting technique, it is possible to reduce the detection efficiency to what is proportional to the total g-ray energy(7). Provided that the weighting function of the g-ray detector is applied to the pulse height spectrum of g-rays, we can obtain the relation Fig. 4 Pulse height spectra for 24-keV neutron capture in PbI2 and 238U samples Measurements were also performed with a sample of carbon in place of other samples and without any sample. These measurements made it possible to determine the time-dependent background under the 24-keV peak in the time-of-flight spectrum. DI. ANALYSIS 1. Principle of Analysis The cross section of neutron capture by a nucleus is usually measured by detection of -rays emitted promptly from an excited state g of the compound nucleus. The g-ray detector ( 2 ) where Wi, is the weighting function for the i-th channel of pulse height spectrum, (BE) the neutron binding energy of the compound nucleus and En the incident neutron energy. Here the chance to detect coincidently two or more g-rays in the same cascade decay is neglected because its probability is very small. The neutron flux can be derived from the measurement on the 10B sample, if the neutron capture probability of the 10B sample and the detection efficiency of the C6F6 detector for the 480-keV g-ray are known. The capture probability of the 10B sample, YB can be estimated from the 10B(n, a) cross section and the calculation of the sample scattering correction. To determine the detection efficiency of the C6F6 detector for the 480-keV g-ray, the counting rate for the r-rays from the 10B sample was compared with those from a standard sample at the neutron energy, where the standard sample had a "black" saturated

Vol. 15, No. 9 (Sep. 1978) 641 capture resonance. In this experiment the 5.2- ev resonance in 109Ag was used as the black resonance. To the g-ray pulse height spectrum for the resonance neutron absorption in the standard sample, the pulse-height weighting was applied again to reduce the detection efficiency proportional to the neutron binding energy of the compound nucleus, 110Ag. From the quantities described above, we obtain a formula for the 24-keV capture probability in a sample, ( 3 ) where superscript g means the quantities at the black resonance, and subscripts S and B show the quantities for the standard sample and for the 10B sample, respectively. Since the ratios of (CW) to (CW)rs and Crb to CB are included in Eq. ( 3 ), the systematic experimental error and the systematic error in the data reduction tend to cancel each other. The factor f represents the ratio of the correction for the g-ray self-absorption in the standard sample to one in the sample. The values of f used in the calculation of the capture probabilities for samples were from 1.00 to 1.11. 2. Analysis of Capture Data The time-dependent background counts under the 24-keV peak are attributed to the prompt g-rays following the capture of scattered neutrons in the collimators and the surrounding materials of the liquid scintillators. The time-dependent background induced by scattered neutrons with sample was determined by subtracting the counts obtained with an open beam from the counts obtained with a carbon sample. After normalization of the scattering probability, this background was subtracted from the capture spectrum, which was of the capture count for each sample except the 232Th and 238U samples. For the 232Th and 238U samples it became 15% of the capture count. The integration of the capture count was performed over the region shown in Fig. 3. This region extends from 19.1 to 27.2 kev and the effective mean enerqv is 23.7 kev. Even if the integral region is extended to the lower energy, the capture cross sections derived in each case agree with each other within the statistical uncertainty. Since all measurements are directly related to the capture probability of the 10B sample, the 10B(n, a) cross section is important for the calculation of Y B. According to the new evaluation of the 10B(n, a) cross section proposed by Sowerby et al(8), it does not vary as 1/vE but falls to a minimum from the 1/vE level in the region of 25 kev. We calculated the average capture probability of the B sample using Sowerby's evaluated cross 10 section and the neutron flux of the region over which the capture counts were integrated. Although the 480-keV g-rays were detected from the 10B(n,a1g) reaction, the ratios of the B(n, a1g) cross section to the 10B(n, a) cross 10 section at the 109Ag resonance energy and the 24 kev are almost constant. Since the these ratios appear as a ratio in the term CrB/CB in Eq. ( 3 ), it cancels out and does not affect the results. The 24-keV neutron capture cross sections were derived from the capture probabilities in the samples by correcting multiple scattering and self-shielding in the samples. The scattering cross sections of nuclei at 24 kev which were necessary in the calculation of the scattering corrections were cited from BNL-325 and ENDF-B/1V. In the case of 127I, the scattering of neutrons from lead was also taken into account because Pbl2 was used as the sample. Since the thickness of samples were sufficiently thin, the analytical method of the correction1(9) was used and the values of the sample scattering correction factor thus obtained are shown in Table 1. Even if there is a little systematic error in the scattering correction, the error may be canceled out because the same calculations were used in both the capture probability of the 10B sample and the capture cross sections of the samples. 3. Uncertainty of Data The statistical errors of the capture probabilities were estimated from the counting statistics of the weighted spectra and the 10B 5

642 J. Nucl. Sci. Technol., spectrum. These values are shown in Table 1. The systematic errors in the measurement are composed of the following factors. (1) Uncertainty of weighting function : 2% To examine the reliability of the weighting function used in this experiment, the hard -ray spectrum from the black 4.9-eV resonance in 197Au and the soft T-ray spectra from g the black 5.2-eV resonance in 109Ag and 4.3-eV resonance in 181Ta were measured. The spectra were multiplied by the weighting function. As the result of the comparison between them, only a 2% discrepancy remains between the hard spectrum and the soft spectra. The details of the examination of the weighting function used in the present experiment are described elsewhere(6). (2) Uncertainty of 10B(n, a) cross section : 2% Since the uncertainty of the 10B(n, a) cross section was estimated to be 2% at 10 kev by Sowerby et al.(8), we assumed the error of the 10B(n, a) cross section to be 2%. (3) Uncertainty of multiple scattering correction : 2% Although the approximate analytical method was used to correct the sample scattering effect, the uncertainty of the correction seemed to be small because the thickness of samples was thin and the correction twice in the opposite direction. was applied (4) Uncertainty of g-ray self-absorption correction in the sample : 2% The g-ray self-absorption in the sample was calculated by the Monte Carlo method. Because the ratio of the correction factor for a sample to one for the standard sample is included in the final results, the error of calculation cancels out. (5) Uncertainty of value of capture probability for 5.2-eV black resonance : 1% The value of the saturated capture probability for the 5.2-eV resonance in 109Ag was calculated by the Monte Carlo technique. It yielded a value of 0.982 with the error of less than 1%. Thus the over-all systematic error in this experiment was estimated to be about 4%. Since natural indium includes 4.28% 113In isotope, whose capture cross section at 24 kev is not well known, we assumed that the capture cross section was almost same as that of 115In. If the cross section of 113In is actually 50% more or less than the assumed value, 115In capture cross section obtained has about 2% uncertainty. This uncertainty was considered in the final estimated error assigned to the capture cross section of 115In. In the case of experiment for thorium, the total neutron flux impinging on Th sample was normalized to that for the reference sample by the counts of a BF3 neutron flux monitor. Therefore, the capture cross section of thorium includes the uncertainty of monitor counts, which was about 5%. IV. RESULTS AND DISCUSSION The present results of the 24-keV capture cross section are shown in Table 2 along with the results of other point cross section measurements. The total errors calculated from the statistical and the systematic errors are about 5-6% except for 232Th and 238U. Chaubey & Sehgal(10) measured neutron activation cross sections using an Sb-Be photoneutron source. All the cross sections were measured relative to 127I, whose cross section was taken equal to 820 mb. Belanova et al.(11) used the spherical transmission method with an Sb-Be source, which was an absolute measurement of the absorption cross sections. While Rimawi & Chrien(2) have measured the 6

Vol. 15, No. 9 (Sep. 1978) 643 Table 2 Experimental results of 24-keV capture cross sections activation cross sections with the Fe-filtered beam derived from the High Flux Beam Reactor at Brookhaven National Laboratory (BNL). The 10B(n a1 g) reaction was used at BNL as the standard cross section for the determination of the neutron flux, which was used also in the present experiment. Although the present value of 115In cross section is in excellent agreement with the results of other point cross section measurements, the present values for 93Nb, 165Ho, 232Th and 238U differ from values of both Chaubey & Sehgal and Belanova et al. The Monte Carlo interpretation for the shell transmission data of uranium reported by Belanova et al. was carried out by Miller & Poenitz(12), resulting in the capture cross section of 495+40 mb for 238U at 24 kev. This revised result agrees very well with the present, Rimawi & Chrien's(2) and Quan et al.'s(4) data as shown in Tables 2 and 3. The present 127I cross section also agrees well with the result of Rimawi & Chrien. Recently, we have measured the neutron capture cross section of 165Ho from 3- to 80- kev region using time-of-flight method. By normalizing the measured capture yield to the 24-keV absolute yield described in this paper, the absolute capture cross section can be obtained. This result shows very good agreement with the data of Asghar et al.(13) and Macklin(14). Since holmium was suggested as a standard of the neutron capture cross section(15), these agreements are valuable. For the capture cross section of 238U, there are three high resolution measurements(16)~(18). Table 3 shows the average values over the energy interval 20-30 kev derived from these measurements and the results of a 24-keV measurement obtained with the monochromatic neutron beam. The discrepancy between them is about ± 10%. The cause of the discrepancy may be mainly the difference in the standard neutron cross section used in each measurement, but for the experiments using the monochromatic beam the difference in the neutron spectrum also influences the result of the capture cross section because there are many resonances in the region in which data are integrated. Spencer & Kaeppeler(19) measured the shape of the 238U capture cross section and showed a significant intermediate structure in the cross section below 100 kev which corresponds to the structure in de Saussure et al.'s data(16). The shapes of these two experiments are in good agreement. When de Saussure et al.'s cross section is folded into the present 24-keV Fe-filtered beam distribution, a value of 575 mb is obtained. So the discrepancy between the present and their result is about 10%. Table 3 Comparison of experimental results for 238U capture cross sections V. CONCLUSION The Fe-filtered beam technique has made possible the measurements of the point capture cross sections with an accuracy of about 7

644 J. Nucl. Sci. Technol., 5%. As a result of comparing with other experimental results, we can find a good agreement with the data recently obtained by the Fe-filtered beam. On the other hand, a discrepancy up to about ± 10% can be seen between all the reported data, which may be caused by difference in the standard cross section used and a systematic error for the treatment of experimental data. In order to make clear the reason of the disagreement between the data, the experimental error must be reduced. For the present type of experiment, the error can be reduced by improving the methods of the pulse-height weighting and of corrections. ACKNOWLEDGMENTS The authors wish to express their deep gratitude to Prof. T. Shibata and Prof. I. Kimura of the Research Reactor Institute, Kyoto University for their support in making these measurements possible. We are also indebted to Dr. M. Okamoto of the Research Laboratory for Nuclear Reactors, Tokyo Institute of Technology for carrying out the chemical analysis of 10B sample. - REFERENCES (1) BLOCK, R.C., FUJITA, Y., KOBAY ASHI, K., 0o- SAKI, T.: J. Nucl. Sci. Technol., 12(1), 1 (1975). (2) RIMAKI, K., CHRIEN, R.E. : Proc. Conf. Nuclear Cross Sections Technology, NBS special publ. 425, p.920 (1975). (3) BLOCK, R.C., KAUSHAL, N.N., HOCKENBURY, R.W. : Nat. Topical Meeting on New Devel. in Reactor Phys. Shielding, CONF-72091, p. 1107 (1972). (4) QUAN, B.L., PENDT, R.H., BLOCK, R.C.: Precision neutron cross section measurements in support of the LMFBR program, COO-2479-10, p. 17 (1976). (5) YAMAMURO, N., Dot, T., HAYASE, T., FUJITA, Y., KOBAYASHI, K., BLOCK, R.C.: Ref. (2), P. 802. (6) YAMAMURO, N., HAYASE, T., Do), T., FUJITA, Y., KOBAYASHI, K., BLOCK, R.C.: Nucl. Instrum. Methods, 133, 531 (1976). (7) MACKLIN, R.L., GIBBONS, J.H. : Phys. Rev., 159,. 1007 (1967). (8) SOWERBY, M.G., PATRICK, B.H., UTTLEY, C.A., DIMENT, K.M. : Proc. Symp. Neutron-Standards and Flux Normalization, AEC Symp. Ser. 23, p. 151 (1970). (9) SCHMITT, H.W. : Sample scattering corrections in neutron beam experiments, ORNL-2883, (1960). (10) CHAUBEY, A.K., SEHGAL, M.L. : Nucl. Phys., 66, 267 (1965). (11) BELANOVA, T.S., VAN'KOV, A.A., MIKHAILUS, F.F., STAVISSKII, Yu. Ya. : J. Nucl. Energy, A/B, 20, 411 (1966). (12) MILLER, L.B., POENITZ, W.P. : Nucl. Sci. Eng., 35, 295 (1969). (13) ASGHAR, M., CHAFFEY, C.M., MOXON, M.C. : Nucl. Phys., A108, 535 (1968). (14) MACKLIN, R.L.: Nucl. Sci. Eng., 59, 231 (1976). (15) CZIRR, J.B., STELTS, M.L. : ibid., 52, 299 (1973). (16) de SAUSSURE, G., SILVER, E.G., PEREZ, R.B., INGLE, R., WEAVER, H.: ibid., 51, 385 (1973). MOXON, M.C. : The neutron capture cross (17) section of 238U in the energy region 0.5 to 100 kev, AERE-R-6074, (1969). (18) FRIESENHAHN, S.J., et al.: Neutron capture cross sections of molybdenum, tantalum, and 238U, GA-10194, (1970). (19) SPENCER, R.R., KAEPPELER, F. : Ref. (2), D. 620. 8