IDENTIFICATION AND QUANTIFICATION OF RADIONUCLIDES IN HISTORICAL WASTE AT ANSTO

Transcription

1 IDENTIFICATION AND QUANTIFICATION OF RADIONUCLIDES IN HISTORICAL WASTE AT ANSTO McOrist G D., Bowles C.J., Fernando K. and Wong R. Australian Nuclear Science and Technology Organisation Australia Abstract Low level radioactive solid waste arising from research and production activities at ANSTO is currently held in storage for final disposal. This waste is packaged in 200L steel drums and is held in a designated radioactive waste storage facility. Most of such drums have been counted in a commercial Canberra Q 2 low level waste assay system to radiologicaly characterise them. The resulting spectra have been evaluated by an experienced gamma-ray spectrometrist to ensure that all significant peaks present have been correctly identified and that the quantity of each resulting radionuclide has been correctly calculated. The validation process highlighted a number of areas for improvement in the original method used to count each drum as well as in the analysis software supplied with the gamma-ray spectrometer. Some of the problems encountered included unidentified and misidentified peaks as well as errors associated with quantifying the respective nuclides. At present, accurate and reliable gamma spectrometry results for well over 5000 drums has been produced. The availability of validated gamma ray spectrometry results of waste has enabled decisive classification waste and subsequent processing. 1. Introduction Low level radioactive solid waste arising from research and production activities over the past 50 years is currently held in on the Australian Nuclear Science and Technology Organisation (ANSTO) site at Lucas Heights. The resulting radioactive waste contains a wide range of radionuclides with half-lives varying from about a year to thousands of years and emits alpha, beta and/or gamma radiation. This waste is handled and processed according to its physical form (solid or liquid) and half-life. Generally, low level radioactive solid waste was stored in red 200L steel drums. This waste usually consisted of paper, tissues, plastics (gloves, bottles, sheeting), metals, mineral material (process waste, grit blast, soil) and glass. Over 6000 drums of low and intermediate level solid waste are currently stored at Lucas Heights. The radioactivity present in these drums has been determined by gamma-ray spectrometry. 2. Gamma-Ray Spectrometry In December 1996 a commercial Canberra Q 2 low level waste assay system was purchased by the Waste Operations Section at ANSTO to determine the identity of and quantify gamma emitting radionuclides present in 200L drums of waste stored at ANSTO. The Canberra Q 2 system consists of a large shielded cabinet in which 200 L drums of waste containing very low levels of radioactivity are placed for scanning [see Figure 1.]. During this process a base plate supporting the drum revolves at a nominal 10 revolutions per minute, exposing the drum to three HP-Ge, co-axial detectors mounted longitudinally within the counting space of the cabinet. Only drums with relatively low total radioactivity were suitable for counting in the Q 2 system. As a rule of thumb, this was indicated by a contact dose rate measurement of less than 50 µsv/h. Each drum was nominally counted for 600 seconds live time. The resulting spectrum was automatically analysed by Canberra Genie 2000 software using empirical efficiency calibrations to determine the concentration of each gamma emitting radionuclide present in the drum. A hard copy of this data was stored in appropriate folders and an electronic version was stored on the counter computer.

2 A closer examination of this data indicated that in many cases identification problems had occurred. In some spectra some of the peaks present had been incorrectly identified and in others peaks actually present had not been identified at all. In others the peaks had been correctly identified but the calculation of the weighted average mean concentration of the nuclide was incorrect. Still in others, high Compton Background at the 2000 kev end of the spectrum indicated another problem. Figure 1: Schematic diagram of the Canberra Q 2 low level waste assay system Some of these errors could be attributed to the following: 3.1 The energy calibration for drums counted between January 1997 and the end of October 1997 had been performed using only Cs-137 and Co-60 calibration sources. Although this meant that radionuclides with gamma-ray energies greater than about 200 kev were correctly identified those with lower energies such as Am-241, Co-57 and Eu-155 were commonly missed. 3.2 The Genie 2000 software had assigned a region of interest (ROI) to an energy where a peak might be expected to be but no statistical peak was present (ie essentially background). The presence of the nuclide and the resulting activity calculation was therefore incorrect. 3.3 Occasionally even when peaks had been correctly identified the concentration of the resulting nuclide was incorrect. One or more of the peaks used for concentration calculation had an interfering peak from another nuclide located near enough for the software to unable to separate. In many cases the presence of other related nuclides such as in uranium and thorium decay chains confirmed that a problem was occurring in these cases. 3.4 The drum had been counted too soon and the resulting spectrum contained nuclides with relatively short half-lives (ie less than about 30 days. One of the more common methods for determining the identity of unknown peaks in a spectrum is to search libraries for nuclides with longer half-lives only. The presence of nuclides with short half-lives significantly complicated this process.

3 3.5 The high Compton Background at the 2000 kev end of the spectrum was regular indicator of a high % deadtime during counting due to the elevated activity of one or more radionuclides in that drum. Some of the resulting problems included peak shift making identification difficult and loss of peak counts. 3. Correcting Spectra Counting Errors Before the spectra could be studied in detail these errors had to be corrected. When opened, each spectrum was evaluated to determine which corrective measures were required. In some cases more than one corrective measure was required. In others, the spectrum indicated that the original count was inappropriate and the drum was scheduled for recounting. When significant low energy peaks had not been identified the spectrum was recalibrated using the energy calibration of a drum counted after October 1997 (ie one that included Am-241, at 59 kev). This enabled these low energy peaks to be correctly identified. In some cases there were sufficient known peaks in the spectrum to allow recalibration to take place without the need of external data. The problem of assigning an ROI where no peak was present was corrected by the choice of appropriate analysis libraries during the actual spectral analysis step [see 4.1]. When the concentration of a particular nuclide was found to be incorrect it was also corrected during spectral analysis. If the spectrum of a drum indicated that it had been counted too soon after packing (ie short half-life nuclides such as Sc-46, Fe-59, Nb-95, Zr-95, etc were present) it was put aside for recounting. In the case of a spectrum where the original count had a high dead time that drum was also scheduled for recounting at a later time. Investigation into the effect of high dead time on quantification of radioactivity by the Q 2 system had recommended that spectra with deadtimes greater than 35% should not be considered for analysis [1]. The spectra of drums above this limit were not processed but set aside for counting on another drum counting system suitable for counting higher activity drums. After these problems had been addressed in the spectra of drums not requiring recounting the gamma-ray spectrometrist was able to correctly identify major peaks present and correctly calculate the concentration of each gamma emitting nuclide present in the drum. 4. Techniques for Analysing Spectra 4.1 Standard methodology The method used for correctly processing (ie analysing or validating) an individual spectrum developed over time and become relatively standardised. This enabled individual Instructions for spectrum calibration and analysis to be written. The original electronic version of each spectrum was recalled and saved on to the computer containing Canberra Genie 2000 software (in Bld 20B). When the recalled spectrum was opened the peaks present were displayed. An option in the spectrum menu allowed a copy of the original printed report to be displayed (Notepad version). Occasionally if the spectrum was relatively simple (eg contained only a few peaks/nuclides) this report could be accepted as correct and the resulting spectrum was transferred to a Spectrum Reviewed folder. However in the majority of cases the spectrum indicated that at least one analysis error that had to be corrected. In many of these the identity and/or concentration of a nuclide known to be present was incorrect. One of the main reasons for this was that the library used for the original analysis contained over 80 individual nuclides. The result was a significantly increased likelihood that a peak could have two or more possible identities that the software could not resolve.

4 In most cases the creation of more appropriate spectra analysis libraries, based on the major component of radioactivity present solved this problem. Some libraries were prepared that contained mainly reactor based nuclides while others consisted of the natural radionuclides of uranium and thorium. Over time about 10 Standard Primary nuclide analysis libraries were created that contained specific nuclides likely to be present in drums containing particular kinds of waste. For example, drums containing waste that had originated from HIFAR (Reactor) related activities contained fission product and activation nuclides such as Cs-137, Eu-152, Eu-154, Cs-134, Sb-125, Ag-108m, and so on. Drums containing mine process waste contained nuclides associated with uranium and thorium progenies such as Pa-234m, Pb-214, Ac-228, Tl-208 etc. An educated choice of nuclide library enabled the gamma-ray spectrometrist to choose that most appropriate library for reanalysis. When the spectrum was reanalysed using the appropriate Standard Library any significant peak(s) that were not identified by the software were found in the Unidentified Peaks section of the analysis printout. As a guide these were indicated by a peak size of 2-3 counts per second (or more) and/or a Peak CPS % Uncertainty of about 3% (or less). The most effective method of identifying unknown peak was the compilation of a database containing commonly unknown or misidentified peaks, which contained nuclide specific information such as peak energy, nuclide, half-life and abundance of unidentified photon peaks of nuclides commonly found after reanalysis. Throughout the spectra analysis process, this database was continually expanded and updated to include other problematic peaks successfully identified by analysing subsequent drum spectra. Table 1 shows a section of this database showing some problematic peaks at low energy levels. When a peak did not appear in this database, there were a number of alternative methods that could be used. For more in-depth searches two other sources of gamma-ray nuclide data were used Nuclear Data Tables in book form [2,3] and computer software search programs [4,5,6]. However although these methods allowed the correct identification of many obscure and unusual nuclides present in waste drums, their use was significantly complicated by a number of factors. For example, the hard copy version [2] lists nuclides that have a peak of almost any abundance at that particular energy. Also in spectrum being studied a peak may span 4 kev (ie 2 kev either side of its centroid). This could mean evaluating 120 or more possible nuclides if the unknown is a relatively lower energy peak. For these reasons it was necessary to eliminate unlikely nuclides based on such criteria as half-life, other energies from the same nuclide and abundance. Table 1 Unknown or misidentified peaks commonly found after reanalysis (section only) Energy (kev) Radionuclide Half-life % Abundance Eu years Co days Eu years Ce days Se days Co days Te-123m 120 days Sb years Ho-166m Ho-166m Pb-212 (natural) Pb-214 (natural) 7.5 Alternately, computer software search programs were used to make the search simpler by allowing the operator to adjust cut-off parameters such as energy range, half-life and abundance. However this search still resulted in a large number of possible nuclides that had to be individually evaluated by the experienced gamma ray spectrometrist. The spectrometrist analysing the spectra needed to be aware of some potential problems when using computer search programs such as incorrectly specifying the cut-off parameters that possibly eliminated the real nuclide and peak libraries that may have had old or incorrect data (eg Pa-234) making nuclide identification impossible.

5 Over time it was possible for gamma-ray spectrometrist to gain familiarity with the energies of many common unknown peaks and the associated nuclide data that may assist in identification. A balanced mix of intuitive guess and the use of hard copy tables and/or computer search programs proved to be the most effective procedure for identifying unknown peaks. Another factor that significantly affected the ability to correctly identify a radionuclide in a waste drum was the concentration of that nuclide. The most common method of confirming the presence of a nuclide was to locate the second and/or third most abundant peak for that nuclide. If the activity was sufficiently low that these were not seen an alternate method was required. In some cases the presence of another nuclide, such as a progeny of uranium or thorium, helped confirm the identity of that nuclide. When peak interference resulted in an incorrect calculation of radionuclide concentration, selectively changing the analysis library to remove the interfering peak was usually successful. For drums suspected of containing an unusually high level of a particular nuclide (such as when the most abundant peaks indicate thousands of kbq of that nuclide) the analysis library was adjusted to contain a larger number of less abundant peaks reducing the problem of peak misidentification. 4.2 Common spectra identification problems During the process of validating over 5000 gamma-ray spectra it was common to find that some peaks (and therefore nuclides) had been incorrectly identified. Some examples were regularly seen in drums after they were processed. In some spectra the 122 kev peak was occasionally misidentified. The radionuclides Eu-152 (121.8 kev) and Co-57 ( kev) were sufficiently close in energy for this to occur. The presence (or absence) of other Eu-152 peaks usually enabled confident identification. In some cases if the level of cyclotron-produced Co- 57 was high enough the kev peak could also be seen (even if not quantified). The presence of a 1274 kev peak usually indicated that Eu-154 (the result of activation of Eu in HIFAR) was present in the waste drum. However this peak could have also resulted from Na-22 (cyclotron produced) being present. The presence (or absence) of other Eu-154 peaks (such as 723 or 123 kev) usually enabled confident identification. The 351 kev peak found in a waste drum spectrum was usually attributed to the most abundant peak of Pb- 214, the naturally occurring progeny in the U-238 decay chain. All Pb-214 (half-life 28.8 minutes) present must be supported by the long half-life parent, Ra-226 (1600 years). If this was the case, another progeny, Bi-214, must also be present. The most abundant peak of Bi-214 occurs at 609 kev and its presence confirms Pb-214. The absence of the 609 kev peak indicated that Bi-211 (from the U-235 chain) was present. In this case the progenies of long half-life Ac-227 (21.8 years) such as Ra-223, Rn-219 or Pb-211 usually confirmed this. 4.3 Some particularly challenging identification examples Although the identification of majority of significant peaks present in gamma spectra enabled the confident identification and concentration of radionuclides present, a few peaks (and therefore nuclides) proved difficult to identify. In these cases significant extra effort was fruitful while in a few others the peaks remained unidentified. Initial efforts to identify the 1592 kev peak proved unsuccessful. However these drums had significant amounts of Th-232 progenies that appeared to be in secular equilibrium. After some concerted searching the most abundant peak of Tl-208 was found to be 2614 kev and that the 1592 kev peak was the resulting double escape peak. Since the counting and analysis software produced a spectrum that only covered the energy range kev the 2614 kev peak was never seen. Two other peaks at and kev were very difficult to identify. The reason was that the only realistic possibility, Bi-213, had a half-life of only 46 minutes. It was only when statistically significant amounts of Fr-221 were also found in some drums that the link with Bi-213 was established. Both are part of the Np-237

6 decay chain and are actually supported by Th-229. Some degree of confirmation came when the resulting concentration values of these two nuclides agreed closely with each other. The 1157 kev peak was also occasionally seen but was unidentified by analysis software. However the more abundant peaks of the most likely nuclides, Ta-182 or Bi-214, were not present. The presence of only very low levels of other nuclides did not provide any clues. The only nuclide that was a realistic possibility, Sc- 44, had a half-life of only 3.9 hours. More detailed study showed that its parent, Ti-44, had a long half-life (63 years) but no gamma emitters and was supporting the Sc-44. When Np-239 and/or Cm-243 were present the resulting gamma-ray spectra were very complex not only as a result of progeny products but also from other radionuclides (eg actinides) present in the same drum. The problem is that both nuclides have very long half-lives and the most abundant peaks in both nuclides not only have almost identical energies but also very similar abundances. Differentiation between these two nuclides is probably only possible using the peaks, 315 and 334 kev that are only present in Np-239. A technique for accurate identification and quantification of these nuclides is currently being developed. The kev peak was found in only one drum that contained relatively low amounts of a few other nuclides. The most likely nuclides such as Eu-152, Co-57 and Eu-154 were found to be not likely since other more abundant peaks in the case of both Eu nuclides were not present. Also this peak was noticeably far enough away from the required energies to be realistically considered. A recent recount of this drum failed to shed any light on the mysterious nuclide. Its current identity remains a bit of a mystery. Still another peak at kev peak has been seen in a couple of drums but its identification is still proving difficult. Further study into this peak is taking place. 5. Conclusions The process of successfully validating the spectra of drums of low-level waste stored at highlighted a number of deficiencies that had occurred during the original count. Some corrective measures were relatively simple while others were significantly complex requiring a large amount of operator time and expertise. In most cases effective spectrum recalibration and other corrective measures resulted in a spectrum that could be successfully processed at a later time. Drums containing short half-life nuclides or high activity were set aside for more appropriate counting at a later time. The process of validating over 5000 drum spectra enabled an effective standardised analysis methodology to be established. The experienced gamma-ray spectrometrists used a combination of techniques to correctly identify and quantify nuclides present. These included books containing nuclear data, computer identification software packages and experience. In many cases the creation of a more appropriate nuclide analysis library resulted in the validation of these spectra. The identity of a couple of peaks proved very challenging. In some cases after significant time and effort these were eventually identified and in some cases quantified. However, even after careful and detailed evaluation of some spectra the confident identification of some peaks still remains a mystery. In conclusion, this process lead to confident identification and quantification of all gamma-emitting nuclides present in the vast majority of drummed low level radiative solid waste stored at ANSTO. The radionuclide information collected will enable accurate classification of waste, and hence facilitate the formulation and selection of conditioning techniques and final disposal options. 6. References 1. R.K.Barnes, C.Bowles, K.Fernando, L.Mokhber-Shahin, D Maher and S.Auld, Review of Gamma Spectral Characterisation Techniques Used on Low Level Solid Waste at ANSTO (WOTD/TR22) 2. Atomic Data and Nuclear Data Tables Vol. 29, No.1, 1983, pp

7 3. Atomic Data and Nuclear Data Tables Vol. 29, No.2, 1983, pp Hacker C., Radiation Decay Version 3.6, May Gamma Ray Spectrum Catalogue 4th edition, September ORTEC Nuclide Navigator Version 3.4, June 2000.