Karin Decker, Colin G. Sanderson and Pamela Greenlaw
Environmental Measurements Laboratory
U. S. Department of Energy
201 Varick Street, 5th Floor
New York, NY 10014-4811
The Gamma Spectrometry Data Validation Program is designed to assess the capability of DOE laboratories and DOE contractors in performing routine gamma spectra analyses which are required for EM projects, site evaluations and other DOE programs. Data reduction of gamma spectra are normally performed with computer codes supplied by commercial manufacturers or are developed "in house". Evaluation of some of the commercially available software was performed at EML in 1987 (Sanderson, 1988), and again in 1991 (Decker and Sanderson, 1992). The first study indicated that there were substantial differences in the ability of the programs to detect small peaks and deconvolute overlapping peaks. The second evaluation showed most of the programs had fairly good results in peak detection and deconvolution, but the analysis of a complex spectrum still gave spurious results. Many of the available commercial programs are easy to use and do not require the user to be an expert in gamma spectrometry. This may lead to errors because even the best programs occasionally give inaccurate results without user intervention.
The participants in the Program receive a data disk or tape containing synthetic spectra designed to test the sensitivity and capability of their HPGe spectrometry reduction system. The participating laboratories are asked to identify and quantify the sample spectra. The spectra are designed to test both the gamma-ray spectrometry software and the ability of the user to properly utilize the software. The Gamma Spectrometry Data Validation Program is sponsored by the Characterization Management Program of the Office of Site Operations (EM-70) within the DOE Office of Environmental Management (EM). The Program is administered as part of the services provided through the Environmental Measurements Laboratory's Quality Assessment Program (EML QAP) for environmental radiological analyses (Sanderson and Greenlaw, 1996). The first distribution for the Program was in May 1996, 31 laboratories participated. In the future, the synthetic spectra will be sent out annually by EML with the results to be reported in 90 days. The expected values will be distributed immediately after the reporting deadline and an evaluation report will be issued by EML in 90 days.
The Gamma Spectrometry Data Validation Program is designed to test the ability of the commercially available software and "in house" programs to accurately identify and quantify the nuclides in complex spectra, independent of the sample geometry or matrix. Synthetic spectra were created using a computer code, SYNTH, developed at the Pacific Northwest National Laboratory and converted into a variety of formats which could be read by most PC based systems. The available formats included APTEC, APTEC Version 4.3, ASCII, CANBERRA Spectran-AT, CANBERRA MicroSAMPO, CANBERRA Gamma-AT, CANBERRA Genie Systems, IAEA Ganaas and Qxas, NUCLEAR DATA Asap, NUCLEUS PCA, ORTEC Minigam II, ORTEC ACE, SILENA SilGamma and SILENA EMCAplus. The data was also put on magnetic tape for users of microvax based systems.
The data used in this evaluation simulates the spectra obtained from air filters counted 10 cm from a 22% coaxial detector with a 0.5 mm beryllium window. The detector is calibrated at 0.5 keV/channel. A calibration spectrum, a background spectrum and three sample spectra of increasing complexity were included for each format. The calibration spectrum contained nuclides covering the energy range from 59.5 keV to 1836 keV. The first two samples contained fallout nuclides with half lives of over 30 days. Naturally occurring nuclides were also present. The third sample contained both short and long lived fission product nuclides. Participants were asked to report values and uncertainties as Becquerel per sample with no decay correction.
The control limits are those established by EML for the Quality Assessment Program (QAP) for various analyte/matrix pairs (Pan, 1995). When available, the analyte/water pair was chosen to simulate the synthetic spectra since there should be no variability among individual water samples. When the analyte/water pair was not available, the analyte/air filter pair was chosen. Participants' performance in QAP is evaluated based on historical analytical capabilities for the individual analyte/matrix pairs. The control limits were established from percentiles of historic data distributions from the last five years of QAP results (10 distributions). The QAP criteria for acceptable performance, "A", was chosen to be between the 15th and 85th percentile of the cumulative normalized distribution, which can be viewed as the middle 70% of all historic measurements. The acceptable with warning criteria, "W", is between the 5th and 15th percentile and between the 85th and 95th percentile. The not acceptable criteria, "N", is established at less than the 5th percentile and greater than the 95th percentile, the outer 10% of the historical data. If there was no historical QAP data for an analyte, the limits were set as: acceptable performance, "A", if the reported value was between 0.8 and 1.2 times the EML value; acceptable with warning, "W", if the reported value was between 0.5 and 0.8 or 1.2 and 1.5 times the EML value; and not acceptable, "N", if the reported value was less than 0.5 or greater than 1.5 times the EML value. These control limits are listed with the EML values. The reported uncertainties may be greater than the acceptance range. For example, historically 70% of the results for 137Cs in water are within 0.97 to 1.15 times the true value which causes some of the reported values here, which are within the counting error of the true value, to fall outside the acceptable range.
The results from the Gamma Spectrometry Data Validation Program are listed by laboratory and software program. All values are Becquerel per sample with no decay correction. The laboratories' results are compared to the experimental value, labeled the EML value, as developed from the synthetic spectra creation program, SYNTH. The ratio is the laboratory value divided by the EML value. The error terms are the uncertainties reported by the laboratories. Under the evaluation column, "A", "W" and "N" are based on the control limits described above. In addition, "FP" indicates false positive and "ND" indicates not detected (false negative). A summary of the evaluations for each sample spectra shows that 62%, 63% and 53%, respectively, of the reported results were evaluated as "acceptable". Only 5-6% of the data results were evaluated as "not acceptable".
Sixteen software packages were evaluated. In general, the results do not appear to be dependent on the software used. Rather the data reflects the ability of the individual laboratories to properly utilize the software.
The results for the first two samples which contained four and six nuclides, respectively, were superior to the third sample which was a more complex spectra containing 15 nuclides. Some of the false positives appear to be due to not correcting for the background using the spectra provided. In sample three, Dysprosium-166 (Dy-166), whose two most abundant lines fall at 46.7 keV and 47.5 keV, was correctly identified by only one laboratory. Many laboratories may have identified this doublet as being the 46.5 keV line of 210Pb.
Decker, K.M. and C.G. Sanderson. "A Reevaluation of Commercial IBM PC Software for the Analysis of Low Level Environmental Gamma-Ray Spectra", Appl. Rad. and Isotopes, 43, 323-337 (1992)
Pan, V. "Analysis of EML QAP Data from 1982-1992: Determination of Operational Criteria and Control Limits for Performance Evaluation Purposes", U.S. DOE Report EML-564 (January 3, 1995)
Sanderson, C.G. "An Evaluation of Commercial IBM PC Software for the Analysis of Low Level Environmental Gamma-Ray Spectra", Environment International, 14, 379-384 (1988)
Sanderson, C. G. and P. Greenlaw. "Semi-Annual Report of the Department of Energy Office of Environmental Management Quality Assessment Program", U.S. DOE Report EML-581 (July 1, 1996)
|Nuclide||Sample||Value||Lower Limit||Lower Middle Limit||Upper Middle Limit||Upper Limit|
Control Limits are established from historical QAP data, where historical data are insufficient, limits of 20% and 50% are applied.
The following are the recommended performance criteria for analysis of environmental analytes: