Counting Results for Acid used to soak KamLAND AmBe Sources




Zelimir Djurcic and Andreas Piepke
University of Alabama
(Status 7/23/2002)




Introduction and Results
In this note we summarize new leak test results for several AmBe sources intended for deployment into KamLAND's liquid scintillator. At this point two AmBe source have been delivered from LSU to the KamLAND site. Both commercial sources were manufactured by Amersham. One of the sources contains 3.33 MBq of 241Am the other 1.85 MBq. In addition UA has shipped another AmBe source to Japan (7/22/2002). These sources will be called A, B and C from here on.
All sources shipped to Kamioka are classified as sealed sources. However, testing any release of activity into a medium surrounding the source has to be done to tight activity limits in order to preserve KamLAND's low background. We have therefore decided to provide an independent leak test. Initial acid soak tests done with both LSU AmBe sources were performed in Japan in May 2002 by Bob Svoboda. The counting of the acid showed clear net activities in both acid samples. This finding has been reported earlier. The sources have been kept in acid for an unknown length of time. It was found that sources A and B released 13.6±1.5 Bq and 1.5±0.3 Bq into the 60 ml of acid used for the soaking, respectively. The soaking was then repeated with water. The water samples were counted and no 241Am activity was found. The activity limits derived for the water samples of sources A and B (60 ml of water, contained in glass bottles, were counted in the same geometry as before) were 0.71±0.52 Bq and 0.49±0.47 Bq, respectively.
It was decided to repeat the initial acid wash in order to understand whether the activity initially removed from the sources was due to a surface contamination or supported by leakage of activity into the solvent. On June 28 both sources A and B were soaked in dilute nitric acid (molarity is not exactly known) for 24 hours. This operation was performed in Japan by Bob Svoboda and Andreas Piepke. The acid samples of both sources and a blank were then shipped to UA for counting. Unfortunately only sample B survived the shipment. In this note we present the results obtained by counting this sample. As in the previous measurement a statistically significant (5.5·Sigma) 241Am activity was found in the wash acid. The measured activity is with 620±160 mBq slightly smaller than that measured before for source B (soaking time 24 hours). Please note that the source had been mechanically cleaned between the two soakings.
In parallel UA's AmBe source has been soaked for 9 days in 0.1 molar HNO3 to provide a backup solution. Results for this source are presented as well. For this sample no 241Am activity was detected. We can place a 90% c.l. limit of 80 mBq of activity released into the acid for this sample. If we assume a constant transfer of activity into the acid the activity limit is 8.8 mBq/day.

Procedure and Data
Am is not well soluble in organic solvents. To provide a stringent leak test the sealed AmBe source to be used in KamLAND have been immersed in 0.1 molar HNO3 which does dissolve Am. The fact that the same source, B, did show a leakage in dilute acid but no leak in water underscores the importance of the right medium (see the previous report).
To obtain a quantitative estimate for the release of activity into the liquid scintillator we would need to scale the measured activities reported here by the ratio, r, of the solubility of Am in scintillator divided by the solubility in dilute nitric acid. r is not known at this time. However, r is for sure less than one. If r is indeed much smaller than unity (as expected) then the acid test provides a stringent test of source leakage. At this point it is not known whether the activity is released at once or is proportional to the exposure time, or a combination of both. It is thus not clear whether it makes sense to divide the activity by the exposure time, as was done in all previous tests.
The acid used in the source soaking tests has been counted at UA using our low background Ge detector. A 59.6 keV gamma is emitted with a branching ratio of b=0.359 gammas/decay in the 241Am alpha decay. The level scheme is shown in the following figure. [link to PDF file] Figure 1: Level scheme of 241Am alpha decay, as taken from the Table of Isotopes, 1998 edition.

In our tests we were looking for a full absorption line at 59.6 keV.
Due to the low energy of the emitted photons, absorption losses in the passive detector materials can severely degrade the detection efficiency. The detection efficiency has been experimentally calibrated for the glass bottles used. We made use of an open Am activity, dissolved in acid. It was found that the combined efficiency is: (3.1±0.16)·10-3 counts/gamma.
[link to PDF file] Figure 2

The following data sets were used in the activity analysis:
*UA Am Be source (C) (blue spectrum)
*LSU 1.85 MBq Am Be source (B) (red spectrum)
*Ge detector background (purple spectrum)
Figure 2 shows the measured energy spectra, normalized to counts per day. The soak sample of source B shows a clear peak at 59.6 keV, the expected energy for 241Am. The soak sample of source C shows no such peak. The fact that the neighboring peaks show more or less the same intensity and widths demonstrates that the data has been taken under the same conditions. The detector background is a bit lower than the soak spectra. This event excess is caused by K/Th/U activities contained in the glass of the bottles. The use of plastic bottles of the same size would allow to obtain tighter activity limits.

The following figures show the different data sets separately, together with a a Gaussian peak fit. [link to PDF file] Figure 3: Source B (weaker LSU source) [link to PDF file] Figure 4: Source C (UA source) [link to PDF file] Figure 5: Detector background

The x-axis is in channel number, the y-axis in counts/channel. Fit parameter 1 corresponds to the peak area (in counts), parameter 2 to the peak centroid (in channels) and parameter 3 to the peak's FWHM (also in channels). The remaining two parameters describe the linear background. For source C and the background no peaks were found. For these fits the peak position and width was fixed to the value obtained in the analysis of source B.
As can be seen from figure 5 a small peak at 59.6 keV (or 173.7 channels) is present. This (2.2·Sigma) net rate has to be subtracted from the source data. It should be noted that it would be better to subtract blank data taken in the presence of the glass bottles. In case of source C this might help to achieve an even tighter activity limit. For source B the blank was lost during transport.
To arrive at the above activities the fitted peak areas were divided by the measuring time, the branching ratio and the detection efficiency.

Conclusion
The AmBe sources provided by LSU show signs of Am leakage. In case of source B this has been demonstrated twice now. Although leakage seems to be the most likely explanation a repeated contamination through the use of contaminated plastic bags (into which the sources were unfortunately put back after the acid treatment) or other materials is not excluded. This can only be shown by yet another round of measurements.
As UA's AmBe source (shipped to Japan) shows no sign of contamination its deployment seems to be the safest and fastest way to provide an energy calibration for KamLAND.

This page is maintained by A. Piepke
Last update July 24, 2002