Experimental Release of 131I: the Green Run

Page 1

Abstract-In December 1949, a large amount of 13'1 was released to the air at Hanford during the dissolving of irradiated uranium fuel for a classified military experiment called the "Green Run." Reports of the release have varied from about 0.15-0.3 PBq. Using, as a guide, the reported measurements of I3'Xe released during the experiment, the amount of I3'I released has been reanalyzed. The results indicate that about 0.40 f 0.12 PBq (11 +. 3 kCi) was released, somewhat larger than the largest previous estimates. Key words: 13'I; contamination, environmental; emissions, atDURINGits operations from 1944 to the present, the Hanford nuclear facilities released radionuclides to the environment. The releases were much larger in the early days before the development of control equipment and restrictive regulatory limits. These releases resulted in radiation doses to the surrounding populations. An extensive study of these doses is currently under way. This study, called the Hanford Environmental Dose Reconstruction Project (HEDRP), is being camed out by the Pacific Northwest Laboratories under the direction of a Technical Steering Panel (TSP). The study, which began in 1988, is a successor to the Hanford Health Effects Study camed out by the state of Washington and is the result of intense public interest and concern about the releases. One of the singular events in the history of the Hanford releases, the Green Run experiment, occurred in December 1949. The experiment was part of the development of a monitoring methodology for intelligence efforts regarding the emerging Soviet nuclear program. The complete details of the experiment remain classified. A declassified description of the experiment has been released (Jenne and Healy 1950). Declassified reports dealing with or mentioning the Green Run show considerable disagreement in the release amount (Jenne and Healy 1950; Paas and Sin- *Department of Environmental Health. School of Public Health and Community Medicine, University of Washington, Seattle, WA ceived 28 January 1992, accepted 24 February 1992) Copyright 0 1992 Health Physics Society glevich 1950; Singlevich 1950; Roberts 1958; Anderson 1974). The values have ranged from 0.15 PBq (4,000 Ci) (Singlevich 1950) to 0.30 PBq (7,780 Ci) (Jenne and Healy 1950). There have been expressions of concern by various members of the public and by representatives of the states of Washington and Oregon about these It is important for the HEDRP that source terms be determined as accurately as possible. Although the earlier estimates indicated that the Green Run release represented less than 2% of the total 13'1 releases from the Hanford facility, a defensible value is required for the HEDRP dosimetry. The purpose of the present study is to provide as good an estimate of the release from the Green Run experiment as is possible from the information contained in the historical record. THE GREEN RUN Examination of the 100 Areas Month1.v Report for November 1949 (Barnard 1949), the P Division Monthly Report for November 1949 (Lee 1949). and the Daily Operating Conditions Report for the 100 tor supplied the irradiated fuel for the Green R u n Experiment. The reactor outage began at 3 a.m. and reached nominal zero power at 4 5 0 a.m. on 16 Novemb e r 1949. Fuel was removed (pushed) from the reactor on 17 November 1949 (Jenne and Healy 1950). Approximately 1,000 of the discharged irradiated fuel slugs (two tons) were made available for the Green Run experiment. The radioactivity in the slugs was allowed to decay for only about 2 wk. The decay time was much shorter than usual to ensure that there would On 1 December 1949 these slugs were put into a clean dissolver at the reprocessing plant. The aluminum jackets were dissolved on the graveyard shift of 2 December (12 midnight to 8 a.m.). The dissolving of the declad and rinsed uranium slugs began at 8 p.m. that same day. The plant filters were bypassed to allow all of the evolved iodine to escape. When a reactor is shut down in a normal way. the control rods are driven into the core. The neutron flux decreases quickly as the reactor goes highly subcritical. There is initially about 6.5% residual power that will slowly decrease with the decay of fission products. For a reactor that has been operating for a long time, the

Page 2

initial shutdown fission product power will decrease to abant- 1% of the fbil reactor power in a little over 3 h. The actual neutronic shutdown of the "F" reaqor can be taken-at 3 a.m. on 16 November 1949. "he cooling time to the start of dissolving is 401 h measured from 3 am.on 16 November 1949 to 8 p.m. oh 2 December Not all of the 133Xeand I3'I were discharged at the same time. The discharge of the xenon was spread out from the start of dissolving over 750 min while the iodine was emitted from about 50 to about 970 min after the start of the dissolving (Jenne and Healy 1950). Jenne and Healy (1950) give the time history of the evolution of these isotopes as the dissolving proceeded. The iodine release rate as a function of time is very symmetric around the midpoint of the discharge. The xenon release rate is nonsymmetric having a very rapid rise to a maximum at about 2.5 h after the start of dissolving followed by an extended decrease over a period of about 10 h (Jenne and Healy 1950). Using a lease gives average values of 4.8 h and 8 h for the average additional cooling from the start of dissolving for xenon and iodine, respectively. The total cooling times for these two isotopes are then 406 h for 133Xe and 409 h for I3'I with an uncertainty of not more than 2 h. The '33Xe available for release from the dissolved fuel comes from the '33Xe in the fuel at the time of reactor shutdown plus the 133Xeproduced by the decay of the 21-h 1331in the fuel at shutdown. About 2.5% of the xenon activity comes from I3'Xe that was produced in the fuel during the cooling period from the decay of I3'I, and its contribution to the ion chamber current in the Jenne and Healy ( 1950) measurement is accounted for. The I3'I available for release during dissolving comes from I3'I in the fuel at reactor shutdown plus the decay of the 1.35-d 13'"'Te in the fuel at shutdown. The contribution from the reservoir isotopes amounts to about 15% additional for xenon and about 5 % additional for iodine over the activities that would be tory at reactor shutdown. In this paper, the estimate of the amount of 8.04d I3'I released from the fuel is based on the amount of 5.243-d '33Xereleased during dissolving as reported by Jenne and Healy ( 1950). All of the contributing isotopes were at radioactive equilibrium at the time the reactor was shut down for fuel discharge. The Bateman (19 10) equations were solved for the equilibrium activities and for the activities of 133Xeand I3'I in the fuel at that time of dissolving. At the time of the Green Run experiment, the reactor fuel was made up of solid slugs of uranium metal clad in aluminum. The average specific power ton-') (Lockwood 1958). It is assumed that at this power density there is little migration of fission products June 1992, Volume 62. Number 6 within the uranium metal and only the fission products formed in the uranium that was actually dissolved are available for release. The amount of iodine discharged from the plant stack is proportional to the assumed value of the fraction evolved from the process solutions. As part of the experiment, samples were taken of the gases and vapors that escaped from the dissolver and were carried up the stack of the reprocessing plant. A pump pulled about 0.472 L S-' (0.5 cfm) through a sampling line at the 15.24-m (SO-ft) level of the stack into the analysis building. The gas stream was passed through a scrubber solution and expelled out the building stack. Samples of the gas stream were collected about every 1 5 min into evacuated 1-L spherical flasks tion (Jenne and Healy 1950). After standing several days, the scrubber solution removed all of the iodine from the sample. After this time, each scrubber flask was connected to an evacuated 1-L spherical ionization chamber made from a boiling flask and the xenon was allowed to divide between the two flasks. The ion chamber was carefully bled to equilibrium with the atmosphere (Jenne and Healy 1950). The ionization current generated in the ion chamber was measured by a vibrating reed electrometer. Jenne and Healy quote Thorburn (1950) who describes the chamber as operated at 90 V in an airconditioned room maintained between 22-24°C and 50% relative humidity. Thorburn (1950) reported constant ionization current per unit activity above 67.5 Jenne and Healy (1950) expressed the release rate of xenon as curies per minute by converting the chamber ion current to flask activity by use of a calibration their calibration factor by assuming that: 1) the average specific ionization of beta particles was 120 ion pairs (ip) per centimeter (ip cm-I); 2) the average path length in the ion chamber was 6 cm; and 3) the ionization from gamma rays was negligible. The results of the ion chamber measurement were reported as the activity discharge rate of '33Xeassuming that: 4) the half-life of 133Xewas 5.4 d and 5 ) that the flow rate up the stack The first assumption appears to be based on applying to the entire '33Xebeta spectrum the specific ionization for an electron whose energy is equal to the average beta energy and assuming that the specific ionization was constant throughout the path of the electrons. Using the average energy of the main beta to determine the energy deposition is not adequate for the following reasons: 1 ) the specific ionization is not linearly proportional to the electron energy; 2) the average energy of conversion and Auger electrons is about onethird that of the beta particles; and 3) the fractional energy deposition of the lower energy discrete electrons is much greater than for the beta particles; more than one-half of the energy deposition from 133Xein the

Page 3

chamber comes from conversion and Auger electrons that Jenne and Healy ( 1950) did not take into account. The second assumption is incorrect. The average uncollided path length to the wall in a spherical volume of radius, R, for the particles emitted unifomly within the volume by an isotropic emitter is 0.75R (Lind the average path length is 4.65 cm. The third assumption is correct. The fourth assumption is slightly higher than the current value of 5.243 d (Walker et al. 1989). For a standing time in the collection flask of as much as 5 d, the difference in the reported discharge rate of '33Xe Even in 1949, air flow rates could be accurately measured, so the fifth assumption is probably reasonable and good to within a few percent. Thorburn (1950) reported ion chamber backgrounds on the order of 2-5 fA for a 2-L chamber. The 1-L chamber used in the Jenne and Healy experiment would have about one-half this background. For the smallest xenon release rate measured by Jenne and assumed stack air-flow rate implies a chamber current of about 1 PA, at least 400 times the background. Measurement of the iodine in the scrubber solution gave a value 2 orders of magnitude lower than expected (Jenne and Healy 1950). They discarded the results and took samples of the condensate in the sampling line. Based on these samples, Jenne and Healy (1950) reported a release of 0.29 PBq of 13'I. No description of the analytical methodology was given. Because Jenne and Healy's lack of methodological detail for obtaining the reported iodine release, this analysis was based on the release of '33Xe. Two models to calculate the energy deposition in a spherical ion chamber were examined. The first was exponential attenuation for the beta particles and the transmission probability model of Kobetich and Katz (1969) for the discrete energy conversion and Auger electrons. The second was based entirely on the Kobetich and Katz model. Evaluation of the two approaches for beta particles was made by comparing the calculated calibration factors with experimental values for a set of ion chamber measurements of "Kr in air. "Kr decays by beta emission with negligible production of conversion or Auger electrons (ICRP 1983). Gamma rays and x rays deposit negligible energy in the ion chambers and were ignored. The measured calibration factors for spherical ion chambers containing "Kr in air were kindly provided by the Victoreen Instrument Co.+ The Victoreen data tion factors for spherical ion chambers containing *'Kr in air. Victoreen, Inc., 10101 Woodland Avenue, Cleveland, OH 44104. Table 1. Experimental calibration factors for spherical ion a Data from Ulman;' values f 20%. are shown in Table 1. Details of the exponential model and the Kobetich and Katz model are given in the Appendix (see also Table 2). The agreement with the Victoreen experiment was much better using the exponential model for the absorption of the beta particles. This model was used in calculating the contribution of the continuous spectrum beta particles to the calibration factor for 133Xe.The exponential model does not apply to electrons emitted with discrete energies. For the discrete energy conversion and Auger electrons emitted in the decay of I3'Xe. absorption fractions were calculated based on the Kobetich and Katz (K-K) transmission probability model. The energy deposition from gamma rays and x rays is very small and was neglected. Evans ( 1955) shows that the exponential attenuation model holds well in the energy range from 1003000 keV. A numerical evaluation of the K-K model was made by comparing the absorption fractions in a 6.2-cm air-filled spherical ion chamber for both allowed and first-forbidden spectrum betas with maximum energes from 100-700 keV. The results showed that the ratio of the absorption fraction from the adjusted exponential model to that given by the K-K model using a first-forbidden spectrum starts out at about 1.3 at the highest energy and decreases rapidly with decreasing energy. At the lowest energy, the ratio had fallen to about 0.75. When an allowed beta spectrum was used in the K-K model, the ratio decreased from I . 1 at the highest energy to about 0.7 at the lowest energy with a minimum (0.65) at 200 keV. An approximate average for the ratio for beta spectra of 100-200 keV maximum energy is 0.7. Further details of this calculation are given in the following uncertainty section. There are large numbers of electrons emitted in these lower-energy spectra with energies similar to those of all of the discrete electrons emitted by I3'Xe and 173Xe.As a result, there is considerable uncertainty in the absorption fractions calculated by the K-K model. For the discrete electrons, their energy depositions were calculated using the K-K model, summed and multiplied by 0.7. The scaled energy deposition was used with a large associated uncertainty. The details of the energy absorption calculation are described in the Ap-

Page 4

490 Health …