Journal of NUCLEAR SCIENCE and TECHNOLOGY, 14[10] pp. 705~706 (October 1977). 705 Application of NaI(T1) Scintillation Detector to Measurement of Tritium Concentration* Kazushige NISHIZAWA, Yoshio ENDO and Mutsuaki SHINAGAWA Department of Nuclear Engineering, Faculty of Engineering, Osaka University** Received September 10, 1976 Revised February 17, 1977 Measurements of tritium concentration in various media have been successfully performed by detecting with an ordinary NaI(T1) scintillation counter the bremsstrahlung emitted from tritium decay. To distinguish meaningful signals from noises emanating mainly from thermoelectrons present in the phototube, differences in rise time were discriminated by means of an electronic circuit, instead of separation by pulse height analysis. This pulse-shape discriminator successfully reduced noise counts below 18 kev in a multichannel analyzer from 100 cpm to 4 cpm, which permitted direct counting of tritiated water as dilute as 1 mci/ml, without requiring any complicated preparation. KEYWORDS: tritium concentration, radiation measurement, bremsstrahlung, windowed detector, pulse shape discriminators, counting circuits I. INTRODUCTION Tritium is a unique nuclide in that it is a soft b-emitter of 18 kev maximum energy and 5.7 kev average, while producing no r-rays. The weak transmissibility of soft b-particles makes it difficult to detect tritium by ordinary windowed radiation counter. For this reason, detectors of liquid scintillation or else gas-flow types are currently used for measuring tritium concentration. Both types of detectors, however, have drawbacks, in that skillful mixing of the tritium-containing samples is required by the scintillator, while the latter form of detector can only measure gaseous samples. Both types, moreover, necessitate troublesome pretreatment procedures for dealing with solid samples, e.g. light metal containing hydrogen. If the bremsstrahlung generated by the P- rays by interaction with a suitable medium could be utilized, the stronger penetration of the resulting radiation should permit the use of an ordinary windowed detector(1) for detection, which would greatly facilitate the measurement of tritium concentration. The difficulty here is the low emission probability of bremsstrahlung and the self shielding effect of the sample, which would normally prevent effective detection of the photons radiating from the surface of the sample. Assuming the true range of 5.7 kev b-particles to be 0.06x 10-3 g/cm2(2), the energy converting to bremsstrahlung is found by calculation to be 2.5x 10-4 kev/b particle in water. This corresponds to 9.25 kev/mci tritium. Consideration solely of the possibility of conversion from b-particles to bremsstrahlung is not adequate, and account must also be taken of self absorption in the source, which would further reduce the intensity of the photons emerging from the surface of the sample. The energy spectrum of the bremsstrahlung emitted by tritium decay has been calculated for tritiated water(3) on the basis of Bethe-Heitler's theory. Since the photon ener- * Work supported in part by a "Grant in Aid for Scientific Research Results" of the Ministry of Education. ** Yamadakami, Suita-shi, Osaka. 17
706 J. Nucl. Sci. Tehcnol., gy of bremsstrahlung cannot exceed the maximum energy of b-particles emitted by tritium decay, the actual spectrum must be shifted further to the lower energy side. This would make it difficult to discriminate the meaningful signals from noise pulses emanating from thermoelectrons present in the detector and from the electronics circuits. An ordinary pulse height discriminator applied to eliminate the noise pulses would concomitantly cut the signal pulses down to useless level. The noise pulses generated by the thermoelectrons in the photoelectron multiplier tube have a rise time appreciably shorter than the scintillation pulses of NaI(T1) crystal(4). This suggested the possibility of applying a pulse shape discriminator to a currently used experimental system, and the present authors examined the utility of this means in effectively lowering the background noise level, with a view to developing a rapid and simple method of measuring the concentration of tritium with use made of an ordinary NaI(T1) scintillation detector. II. EXPERIMENTAL A NaI(T1) scintillation detector from Silice and Quartz Co. Model 2S5B1, embodying a 25 mm dia. 2 mm thick crystal with Be window and 10-stages photoelectron multiplier, was used in combination with a low-noise preamplifier. A charge of 1,000 V was applied to the electron multiplier, as specified by the maker. The sample solution was loaded in a counting cup provided with a 5 mm thick polypropylene bottom, and was set above the crystal window as shown in Fig. 1, for counting. The sample-detector system was shielded from high energy g-rays by a Pb wall 50 mm thick. Fig. 1 Sample-detector arrangement Tritiated water of concentration certified to be 1 Ci/m/ in Oct. 1969 by the Japan Isotope Association was diluted to concentrations ranging from 100 nci/ml to 100 mci/ml by adding distilled water. Prior the experiment, a number of these samples were calibrated against standard tritium using a liquid scintillation counter (Model LSCI), both from Nuclear Enterprise. One milliliter aliquots were made of solutions of each concentration, for counting with the bremsstrahlung method. The measuring system is schematized in Fig. 2. The entire circuit is made up of NIM components, with the exception of the pre-amplifier. Fig. 2 Block diagram of tritium bremsstrahlung counting system 18
Vol. 14, No. 10 (Oct. 1977) 707 The output pulses from the double delay line amplifier (Canberra Model 1141) are fed into the two constant-fraction timing single channel analyzers (Ortec Model 455), one set to 0.1 fraction rate, and the other to 0.5. The timing signals from the two analyzers provide the time to pulse-height converter (Ortec Model 451) with start and stop triggers, respectively. Thus the lapses of time between fraction rates 0.1 and 0.5 were converted to corresponding pulse heights. Figure 3 shows the pulse height distributions obtained with rise time intervals of 50 and 100 nsec after the pulser signal ; it is seen that discrimination of rise time between 50 and 100 nsec is quite possible. The pulses thus identified to be bremsstrahlung signals emanating from tritium are led to the gate input of the multi-channel analyzer, for slow coincidence measurement with the analogue pulses of ordinary input. phototube noises being found most prevalent in the low voltage regions of pulse height. The background counts below 13 kev amounted to 100 cpm. The actually measured difference in rise time between the tritium bremsstrahlung signals and the background noise pulses is clearly revealed in Fig. 4, which represents the time spectra obtained through the timeto-amplitude converter. The bremsstrahlung is seen to rise to its peak only at the 108th channel, by which point in time, the background has already lowered quite significantly from its highest value. This background spectrum is constituted by noise pulses resulting not only from the electron multipler thermoelectrons already mentioned, but also from the electronics circuits, from cosmic rays and from scattered radiation of high energy g-rays. Fig. 3 Characteristics of rise time discriminator III. RESULTS AND DISCUSSION As expected, an ordinary pulse height discriminator trially connected to the present experimental system for bremsstrahlung proved incapable of effectively discriminating signals representing tritium bremsstrahlung- with its energy spectrum peak at 5.8 kev-from background noises generated by thermo-electrons in the photo-electron multiplier, the Fig. 4 Rise time distributions of background and bremsstrahlung pulses of the NaI (T1) detector In order to determine the tritium concentration in the tritiated water, the interval indicated by the shaded zone in Fig. 4- representing a range of time lapse extending from 230 to 270 nsec-was set on the single channel pulse height analyzer for opening the gate of the slow coincidence circuit. This time discrimination reduced the background counts from 100 to 4 cpm with the present experimental system, at a sacrifice of only about 19
708 J. Nucl. Sci. Technol., 10% in the counting efficiency for bremsstrahlung signals, thus definitely confirming the effectiveness of the rise-time discriminator in removing noise pulses from a scintillation measuring system, which should prove particularly useful for counting low-energy radiation such as the bremsquanta of tritium. The observable bremsstrahlung intensity is not only influenced by the differences in its emission probability, which depends on the medium in which tritium is contained, but also by the absorption coefficient of the medium through it travels, and which varies with the atomic number. Hence the counting rate obtained on samples containing tritium would vary from substance to substance constituting the sample. Figure 5 shows examples of data on counting rates for bremsstrahlung radiating through methyl alcohol (99.5v/0), through water and through lead acetate solution. The duration of counting was 1,000 sec for all three media. For the water sample containing 1 mci / ml tritium, the net rate was about 20 counts/1,000 sec. Since fluctuations appearing in the background counts were lower than ±10 counts/1,000 sec, tritium concentrations of the order of 1 mci/ml of water could be measured to yield meaningful data. The order among the three different media for the amounts of tritium bremsquanta issuing therefrom, as detected by NaI(T1) crystal, was methyl alcohol - water - lead acetate solution. The loss of electron energy due to bremsstrahlung production in such a case is, according to the Bethe-Heitler theory, proportional to n S xi(z x=1 2 i+ zi)/ai, where zi : atomic number, A,: atomic mass, and x, : weight fraction of the atomic species in question for the multicomponent system ; hence the electron energy loss is roughly proportional to the mean atomic number. The values of x(z2+z)1/a were 2.58, 2.83 and 4.41 for methyl alcohol, water and lead acetate solution, respectively. In reference to the value for water, the energy dissipation of b-particles due to bremsstrahlung in methyl alcohol would be equivalent to 0.912, and to 1.56 in lead acetate solution. Radiation loss due to photoelectric absorption in such media is proportional to the fifth power of the atomic number. The coefficients of mass absorption of X-rays of 2.0 A (6.2 kev) are 5.96, 8.35 and 19.26 cm2/g ), respectively, for methyl alcohol, water (5 and lead acetate solution. The correction factors to account for self absorption by these media when determining the tritium content present therein can be obtained by the formula(3) Fig. 5 Effect of differences in atomic number of media on bremsstrahlung counting rate where I/I0 is the correction factor of self absorption, m the mass absorption coefficient (cm2/g) and m the thickness of absorbing layer (g/cm2). The values of I/I0 for methyl alcohol, water and lead acetate solution were 0.153, 0.0234 and 0.0082, respectively, in the case of a 1 ml sample containing tritium placed in a 19.6 mm dia. counting cup. Calculation of the bremsstrahlung and of the self absorption by the sample provided the theoretical intensity of radiation from 20
Vol. 14, No. 10 (Oct. 1977) 709 the surface of methyl alcohol to be 6 times that from water, the corresponding factor being 0.54 for lead acetate solution. The bremsstrahlung counting rate actually obtained from sample media containing a given common tritium concentration was only 2.5 times higher with alcohol than with pure water, while being in the case of lead acetate solution (33.3% Pb(CH3COOH2)), as indicated in Fig. 5. This discrepancy between theoretical and experimental yields must be attributed to underestimation of the absorption by methyl alcohol of the softer radiations in the spectrum, the bremsquanta emitted by methyl alcohol being softer than that from water, whereas the absorption was calculated for a given energy of 6.2 kev. Another factor to be considered is that such radiation emerging from media of low atomic number is absorbed to significant extent by the polypropylene window of the cell and by the Be window of the detector. The foregoing observations would suggest the possibility of obtaining a high intensity of radiation from the surface of a light metal such as Li. While the relative yield of bremsstrahlung in Li amounts to only 0.60 of that in water, this medium conversely offers 13 times higher probability of escape for the bremsstrahlung. Hence, tritium contained in Li metal should be easily measurable by means of the present system. IV. CONCLUSION The bremsstrahlung of tritium can be effectively measured with windowed NaI(T1) scintillation detector, a common tool for measuring low-energy X-rays. For this purpose, the pulse shape discriminator described here effectively reduces the noise pulses generated in the phototube. The concentration of tritium in the various media can be easily measured, without complicated preparation required for the samples. Since the windowed detector can be brought to the sample to be measured, this type of instrument is very suitable for application as portable tritium monitor, as compared with the currently used equipment, which requires the samples to be placed in, or to be mixed in the detector. The fact that the counting efficiency of tritium is higher in media of lower atomic number would indicate this bremsstrahlung method to be particularly suitable for measuring the concentration of tritium present in light metal, e.g. Li. ACKNOWLEDGMENT The authors wish to thank Prof. H. Kakihana of the Research Laboratory of Nuclear Reactor, Tokyo Institute of Technology, for his encouragement and interest in this work. (Text edited grammatically REFERENCES- by Mr. M. Yoshida.) (1) CUTIS, M.L., BENTZ, L.L.: MLM-1369, (1967). (2) WESTERMARK, T., et al.: Nucl. Instrum. Methods, 9, 141 (1960). (3) CAMERON, J.F., et al.: AERE-R3086, (1963). (4) LANDIS, D., GOULDING, F.S.: Nucl. Instrum. Methods, 33, 303 (1965). (5) CLARK, G.L.: "Handbook of X-Rays", (KAELBLE, E.F., ed.), p. 1~6 (1967), McGraw-Hill, New York. 21