- Email: [email protected]
Long-term neutron dose fading study of LiF: Mg,Cu,P TLD
Radiation Measurements 46 (2011) 199e204
Contents lists available at ScienceDirect
Radiation Measurements journal homepage: www.elsevier.com/locate/radmeas
Long-term neutron dose fading study of LiF: Mg,Cu,P TLD J.A. Delzer a, A. Romanyukha b, *, L.A. Benevides b a b
Uniformed Services University of the Health Sciences, Bethesda, MD 20814, USA Naval Dosimetry Center, 8901 Wisconsin Avenue, Bethesda, MD 20889, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 5 October 2009 Received in revised form 5 March 2010 Accepted 10 November 2010
An extensive study of neutron dose fading in LiF:Mg,Cu,P thermoluminescent dosimeters (TLD) using commercially-available dosimeters and equipment for their processing was undertaken. During the 52 week study, three thousand TLDs were stored for various lengths of time before and after being exposed to a plutonium beryllium radiation source. The TLDs were subsequently processed and the resulting doses were compared to the reference exposure. Both a loss of signal and a loss of sensitivity were evaluated. The results of this study have shown that the commercially produced LiF:Mg,Cu,P TLD has no statistically significant change in sensitivity or change in signal with up to 52 weeks of pre-irradiation or post-irradiation time. The results of this study will provide the technical basis for increasing the exposure record accuracy, and extending the usable lifetime of dosimeters. Published by Elsevier Ltd.
Keywords: TLD Dosimetry Neutron Dose fading TLD-600H TLD-700H
1. Introduction Since 2002, U.S. Navy has used the dosimetric system based on LiF:Mg,Cu,P thermoluminescence dosimeters (TLD’s) and Harshaw model 8800 dose readers, developed and produced by Thermo Fisher Scientific (for more details, see Cassata et al., 2002; Delzer et al., 2008; Moscovitch, 1999; Moscovitch et al., 2006). The dosimeter consists of four LiF:Mg,Cu,P elements, mounted on an aluminum substrates which were placed in a plastic holder with the appropriate filtration. Three elements, TLD-700H (7LiF:Mg,Cu,P), are sensitive only to photon and beta irradiation, whereas the fourth element of TLD-600H (6LiF:Mg,Cu,P), is sensitive to photon, beta irradiation and thermal neutron resulting from albedo irradiation. The neutron dose measurement by TLD-600H is based on the following nuclear reaction. 6
Li þ 1 n/3 H þ 4 aþ
(1)
Thus the neutron dose reading is caused by energy deposition of H and 4aþ ions in TLD-600H which is also sensitive to photon and beta irradiation. The neutron dose is calculated as the by taking the difference of the element 4 and element 1 responses. From reaction (1), one can see that as result of each neutron exposure the amount of neutron-sensitive 6Li [(sensitive to neutron)] in element 4 3
* Corresponding author. Tel.: þ1 301 295 4149; fax: þ1 301 295 4165. E-mail address: [email protected] (A. Romanyukha). 1350-4487/$ e see front matter Published by Elsevier Ltd. doi:10.1016/j.radmeas.2010.11.012
steadily decreases. To make things more complex, the tritium generated by neutron exposure decays to 3He with emission of a beta particle (average energy of 5.7 keV) and antineutrino. Halflife of tritium is 12.33 years. Thus, besides the reduction of sensitivity (6Li reduction in element 4), neutron irradiation also results in self-irradiation of the element 4 and the generation of gaseous 3 He. Plus neutron dose determination as difference of the doses of element 1 and element 4 can be affected by possible photon dose fading in element 1. Clearly these facts make it important to study the effects of neutron pre-irradiation fade (sensitivity) and neutron post-irradiation (signal) fade. Very few studies on the neutron preand post-irradiation fading properties of this material have been performed (see Jones and Stokes, 2007; Gilvin, 2007). Most detailed previous study of neutron dose fading (Jones and Stokes, 2007) had established that the main dosimetric signal for this material was stable over a storage time of about 100 days and concluded that the pre-irradiation and post-irradiation fading of a commercially available LiF:Mg,Cu,P TLD irradiated in a mixed photon-neutron field at various storage temperatures were relatively small over a 24 week period. The maximum loss of sensitivity to neutron exposure and neutron signal did not exceed 8% at room temperature at the high storage temperature of 50 C the loss was found 18%. This study was performed with only 40 dosimeters at each of three storage temperature, e.g. 5 dosimeters per one time point, e.g. 0, 7, 14, 28, 56, 84 and 164 days. In difference from our results for photon dose fading in TLD-700H (Delzer et al., 2008; Jones and Stokes, 2007) found a significant increase in sensitivity after 60 days of storage at room temperature. Therefore it is important to clarify
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204 500
80000 70000
400 60000 50000
300
40000 200
30000
Temprerature, C
that observed effects statistically significant at bigger numbers of dosimeters and time points. Present paper represents the neutron portion of a two-phase study that evaluated the photon (Delzer et al., 2008) and neutron fading characteristics of the LiF:Mg,Cu,P TLD (Harshaw Model 8840/41). The methods used for this study were consistent with the previous published methods (Delzer et al., 2008). The main purpose was to determine if the LiF:Mg,Cu,P TLD has a statistically significant loss of sensitivity or loss of signal with either neutron pre-irradiation or neutron post-irradiation fade over a period of 52 weeks on 15 dosimeters for each time point.
TL response, arbitrary units
200
20000 100 10000
2. Materials and methods 0
The TLD, Harshaw Model 8840/41, consists of the TLD card and holder. The card contains four TeflonÒ encapsulated LiF:Mg,Cu,P elements. The holder contains tin, copper, Mylar and plastic filters that allow for discrimination of gamma radiation energy levels and determination of beta and neutron dose information (Horowitz and Horowitz, 1992). The TLD reader used for this study was the traditional Harshaw Model 8800 TLD Reader. A reader calibration factor (RCF) was determined each day, prior to processing any TLDs. Variations in measurements in the TLD reader are caused by environmental changes and are corrected by the RCF. Proper calibration of a reader and dosimeters used in this study is a critically important because of significant length of the study. Details of the used calibration procedure can be found in Delzer et al. (2008). All TLDs were annealed and processed using a time-temperature profile (TTP) that was designed to maximize the sensitivity and reusability of the DT-702. The TTP began with an initial preheat of the TLD to 165 C for 10 s to eliminate low temperature traps that are confounding and short lived compared to the main dosimetric peak (Budzanowski, 2002). This was followed by a 16.67 s acquisition time, where heat increased at a rate of 15 C/s to a maximum of 260 C. The TL response was determined by glow curve integration. Maximum temperatures were limited to 260 C to prevent damage to the TLD material from overheating (Cassata et al., 2002; Tang et al., 2002). A typical observed glow curve and TTP are shown on Fig. 1. No additional thermal treatments were used in the study. Three thousand TLDs were used in this study. The entire set of TLDs were annealed one day prior to the beginning of Day 0 of the study by performing two sequential 2-mrem (20mSv) anneals. This ensured that any residual signal was reduced to the lowest degree possible with the TTP used (Horowitz and Horowitz, 1992; Ramlo et al., 2007). The TLDs were sorted into several sets according to a pre-established irradiation schedule (Fig. 2). Once the sets of TLDs were established, irradiations were performed to emulate normal
0
50
100
400 TLDs stored with fade groups
Post Irradiation Fade 500 TLDs Irradiated to 4 mSv using PuBe Neutron Source Day 0
0 200
Channel Fig. 1. Solid line shows typical observed glow curve. Dash line shows applied timetemperature profile.
pre- and post-irradiation times that would be expected during normal handling. The first TLDs irradiated were the post-irradiation set, done on Day 0. This group was then used for the rest of the study for the Post-Irradiation fade group. The Pre-Irradiation TLD group was irradiated each week. Prior to processing, the Combined Pre/ Post Irradiation TLDs were irradiated to allow for mixed pre- and post-irradiation times. To account for any machine variance, a set of reference TLDs was utilized. This set consisted of a group of 15 TLDs and were processed on Day 0. Following Day 0, the reference set was irradiated along with the Pre-Irradiation TLDs and processed with them. Although these TLDs were annealed prior to irradiating, an additional annealing process was performed for consistency in study design to ensure there was no residual signal on the TLD. Neutron irradiation utilized a plutonium beryllium (PuBe) source and a carousel irradiation method developed specifically for this study. The developed irradiator was able to provide highlyinformed neutron exposure to 100 dosimeters simultaneously. The neutron source consisted of an encapsulated plutonium beryllium neutron source of 76.37 g of plutonium 239 (5 Ci) with a cylindrical shape measuring 3.30 cm (outside diameter) by 8.64 cm high and producing 2.3 106 neutrons per cm2-second per Curie. The source is surrounded by a virgin polyethylene shield to allow for thermalization of neutrons. Borated polyethylene is used to prevent escape of neutrons from the shielding. The delivered dose rate for the irradiation method used was 67 mSv per hour. Using a carousel irradiator, 100 TLDs were mounted on the carousel and evenly distributed keeping Element 4 of the TLD located on the bottom-
3000 TLDs annealed at start of study
Control TLDs
150
Pre Irradiation Fade 500 TLDs Irradiated to 4 mSv using PuBe Neutron Source Just prior to reading
Fig. 2. Establishment of TLD Groups used for fade study.
Mixed Pre and Post Fade 1600 TLDs Irradiated to 4 mSv using PuBe Neutron Source and stored for varying times
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204
201
Fig. 3. a) The top view of the carousel used to irradiate the TLDs. (b) A cut-away view shows the position of the PuBe source in relation to the TLDs cards. It is important to note that the carousel was placed in motion prior to inserting into source and remained rotating after the irradiation ensuring a homogeneous irradiation.
inner radius to be irradiated as indicated in Fig. 3. The TLDs were irradiated to 4 mSv. Rotation of the carousel was initiated prior to irradiation, and continued until the carousel was removed from the PuBe source. The TLDs were removed from the carousel assembly and prepared for processing. 3. Results and discussion Throughout the course of the study, the average of the group was determined, and then evaluated for outliers. The contribution of the control TLDs was subtracted and the data was normalized to the baseline data determined at the start of the study. It should also be noted that the propagation of error was carried through for the entire study. The average dose of each element for each set of TLDs was determined for each data point and the associated standard deviation of each set of data was calculated as the standard error. After processing each set of TLDs, the results of each element were evaluated for outliers to identify any statistically spurious data points using Chauvenet’s criterion (Taylor, 1997). This process was performed once for each set of TLDs (both background and fade groups) obtaining a final mean value and standard error for each data point. After evaluating each set of TLDs for outliers, the mean of the control TLDs was subtracted to give a background corrected reading for each element of the TLD. Standard error propagation methods were utilized as described by Knoll (2000). After correcting each data point for background, the data was normalized against the week 0 TLD results or baseline data, by taking a ratio between Week 0 and Week X. This provided a relative response of signal or sensitivity based on a group of TLDs with no associated pre-exposure or post-exposure fade. The evaluation of the resulting data focused primarily on the determination of fading over time, specifically whether a statistical difference existed in the change of signal and sensitivity for pre, post, and combination of fade for the entire 52 week study. In order to perform this evaluation, some assumptions were made on the specific dosimeter characteristics that included the statistical distribution and the expected variations for the TLD. The test for significant differences was set at a 95% confidence interval and performed among sample means (m) of the independent variable (known signal applied to TLD), and the dependent variable (time). The statistical single element variation in the dosimeter has been reported to be 2e4% (Moscovitch, 1999). In this study, a value of 3.5% was used for reproducibility based on the historical average observed by the Naval Dosimetry Center. The variation was
determined by repeated measurements of the same card with the same dose. Given that the distribution of dose is Gaussian (normal) in nature for the entire population of TLDs, it should be expected that 99% of the readings will fall within three standard deviations (3s) of the average. Therefore, a maximum 3s experimental error of about 10.5% would be expected. While this was not mathematically accounted for in this study, it can be assumed that differences of about 10.5% or less are well within the experimental error. One-way analysis of variance (ANOVA) tests were performed on the results of each element for all variations of storage time. For all ANOVA calculations, an alpha level of 0.05 was used. In the event that a statistical difference was detected by the ANOVA, a multiple comparison test using the Holm-Sidak method was performed (Systat, 2004a). The Holm-Sidak test is a non-parametric test that is more powerful than the standard ANOVA. The test attempts to identify the specific points that are statistically different compared to the control value (no storage time baseline response) (Systat, 2004a). When performing the test, the P values of all comparisons are computed and ordered from smallest to largest. Each P value is then compared to a critical level against a 95% significance level, the rank of the P value, and the total number of comparisons made. A P value less than the critical level (0.05) indicates there is a significant difference between the corresponding two groups. Throughout the study, control TLDs were used to account for background radiation, present from natural and manmade sources. Environmental radiation accounts for approximately 1 mSv per year of the total annual effective dose equivalent in the United States of approximately 3.6 mSv per year (Eisenbud and Gesell, 1997). In this study, the TLDs were stored in an area of much lower background resulting in a daily rate of 1.124 mSv/day or 0.409 mSv/year. Study TLDs were processed weekly and background radiation levels were subtracted from the study groups. In analyzing the background level, two analyses were performed; determination of a consistent daily background rate, and statistical difference between the different elements for the same groups of TLDs across the entire study period. In theory, the rate should remain constant over time and since the contribution to background is primarily from photons, all four elements should have the same background over time. To analyze these data points, a simple linear regression was used to evaluate the daily rate and a Pearson Product-Moment Correlation was used to compare the correlation of the first element with the other three (Systat, 2004b). The Pearson Correlation showed no statistical difference between all 4 elements when individual elements were compared against each other.
202
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204
Average Background Accumulation 450
Background Radiation Accumulation (µSv)
400
350
300
250
200
150
100
y = 7.3077x + 6.2041 R2 = 0.9982
50
0 0
10
20
30
40
50
Week Fig. 4. The graph shows the average of all elements background radiation accumulation. Linear regression is shown as a solid line through the data points. The error bars indicate the propagated errors for the mathematical manipulation (1s). Equation of the regression analysis and R2 value for linear regression fit are listed.
The linear regression for each element resulted in an average R2 value of 0.999. Fig. 4 shows the regression plot, associated equation and R2 value. For the study, the primary focus was to determine the effects of fading due to neutron irradiation, based on results from Element 4. A One-way ANOVA test performed on the Pre-Irradiation TLD group for the combined photon and neutron contributions indicated that there was no significant difference between the responses of the elements for the entire 52 week period. (P ¼ 0.707). The maximum increase was measured at 5 weeks at 5%. The TLD response for sensitivity appeared to remain consistent for entire study period, with a maximum decreases measured at only 6%. To analyze the pure neutron component of pre-irradiation fade, the photon component from Element 1 must be subtracted from Element 4. In order to ensure the geometry was consistent, a study was performed to account for the differences in irradiation position. This was performed by rotating the TLD card through all the positions of the carousel and determining the ratio between position 1 and position 4 for pure photon. In order to apply a correction factor, it must be determined that there is truly a significant difference between the two values due to geometry. This was accomplished by performing a t-test for Element 1 geometry positions (Systat, 2004b). The result of this test identified a significant difference between the two positions (P 0.001). Therefore, when calculating a ratio, a difference of 8% due to the geometry differences that exist between Element 1 and Element 4. After the correction for Element 1 was determined, it was subtracted from the value of Element 4. This methodology was also used by Jones and Stokes, 2007 in determining the pure neutron effects. After performing the neutron calculation, the effects of fade from pure neutron were analyzed. A One-way ANOVA test performed on the Pre-Irradiation TLD group for the neutron contributions indicated that there was no statistically significant difference between the responses of the elements for the entire 52 week period. (P ¼ 0.754). The maximum increase was measured at 5 weeks at 5%. The TLD response appeared to remain consistent for entire study period, with a maximum decreases measured at only 6%. Changes in sensitivity for Element 4 for only neutron are displayed in Fig. 5. An interesting finding in analyzing the data is that
there is little difference between pre-irradiation fade for photon and neutron irradiation. This is not surprising, since the contribution from the photon component was only 4% of the total signal. For that reason, the data provides a minimal insight to the differences of photon and neutron fading characteristics. Element 4 was analyzed for Post-Irradiation Fade, which is considered a change of signal over time. A One-way ANOVA test performed on the Post-Irradiation TLD group for the combined photon and neutron contributions indicated that there was no significant difference between the responses of the elements for the entire 52 week period (P ¼ 0.312). The maximum increase was measured at 5 weeks at 5%. The TLD response appeared to remain consistent for the entire study, with maximum decreases measured at only 2%. After performing the same correction for neutron contribution as used for Pre-Irradiation Fade, the effects of fade from pure neutron were analyzed. A One-way ANOVA test performed on the Post-Irradiation TLD group for the neutron contributions indicated that there was no significant difference between the responses of the elements for the entire 52 week period. (P ¼ 0.766). The
Fig. 5. The plot displays the trend for Neutron pre fade normalized response over a 52 week period. The error bars indicate the propagated errors for the mathematical manipulation (1s).
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204
Fig. 6. The plot displays the trend for Neutron post fade normalized response over a 52 week period. The error bars indicate the propagated errors for the mathematical manipulation (1s).
Fig. 7. The plot displays the trend for equal time between pre- and post- neutron fade normalized response over a 52 week period. The error bars indicate the propagated errors for the mathematical manipulation (1s).
maximum increase was measured at 39 weeks at 6%. The TLD response appeared to remain consistent for entire study period, with a maximum decreases measured at only 3%. Changes in signal for Element 4 for only neutron are displayed in Fig. 6.
203
Fig. 8. The plot displays the trend for the combined pure pre- and post- neutron fade normalized response over a 52 week period. The error bars indicate the propagated errors for the mathematical manipulation (1s).
An analysis was performed on Combined Fade results. Several different combinations are presented with varying results in the data. First, an analysis was done on the combined pure Pre and Post Fade data, then an analysis was performed on equal time between Pre and Post data, and finally an analysis was performed on all of the data. Analyzing the pure Pre and Post results for combined neutron fade with one-way ANOVA resulted in no statistically significant difference (P ¼ 0.860). The results can be seen in Fig. 7. The second analysis performed for the combined fade was determining the effects of equal time between pre and post-exposure time. This group consists of all TLDs that had an equal amount of time between the anneal-expose-read periods. One-way ANOVA initially detected a significant difference (P ¼ 0.008), but HolmSidak testing determined no point to be statistically significant (P ¼ 0.0114). The neutron equal pre and post fade results can be seen in Fig. 8. Finally, a statistical analysis was performed on the entire population of pre and post-exposure fade data. One-way ANOVA initially detected a significant difference (P ¼ 0.006), but the nonparametric Holm-Sidak testing determined no point to be statistically significant (P ¼ 0.0124). This result is quite remarkable considering there are 237 data points, accounting for 2474
Fig. 9. Time dependence of combined pre- and post- neutron dose fade normalized response.
204
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204
TLDs for all the fade groups, over the entire 52 week period (see Fig. 9). 4. Conclusions The LiF:Mg,Cu,P thermoluminescence dosimeters have been shown to have no significant change in sensitivity or signal response with up to 52 weeks of neutron pre-exposure, postexposure, or combination of pre-exposure and post-exposure time. We have not observed previously reported in Jones and Stokes (2007); Gilvin, 2007 variations in both sensitivity (at 60 days storage) and signal for neutron and photon exposures. The differences in conclusions between this study and previous ones (Jones and Stokes, 2007; Gilvin, 2007) can be attributed to the greater statistical power that is obtained with the use of so many dosimeters. Currently the TLDs are normally issued by U.S. Navy for eightweek periods, with up to 12-week maximum for special circumstances. The data presented in this study suggest that extending beyond these limits may be warranted. With this increase, flexibility with international and domestic shipping procedures, as well as reduced workload requirements for dosimetry processing could be realized. When evaluating issue periods, environmental conditions should coincide with the ones established by this study design. This extension of issue periods must be concurrently balanced with the effect of lower limit of detection (LLD), since an increase in LLD may be seen with longer issue periods (see for detail Traino et al., 1998; Romanyukha et al., 2008). Acknowledgements The study was funded through U.S. Department of Defense operational and maintenance budget. JAD would like to give special thanks to the following people for their contributions provided during this research: Dr. Tomoko Hooper, Uniformed Services University of the Health Sciences, for her honest feedback and support throughout. CDR Russ Lawry, Uniformed Services University of the Health Sciences, for his mentorship and questioning attitude throughout the study,
Mr. Dave King, Naval Dosimetry Center, for his contributions in the study design and technical support. The views expressed in this paper are those of the authors and do not reflect the official policy or position of the Navy and Marine Corps Public Health Center, Navy Bureau of Medicine and Surgery, Department of the Navy, Department of Defense or the U.S. Government. References Budzanowski, M., 2002. The Influence of post-exposure heating on the Stability of MCP-N (LiF:Mg, Cu, P) TL Detectors. Radiat. Prot. Dosimetry 101 (1e4), 257e260. Cassata, J.R., Moscovitch, M., Rotunda, J.E., Velbeck, K.J., 2002. A New Paradigm in Personal dosimetry using LiF: Mg, Cu, P. Radiat. Prot. Dosimetry 101, 27e42. Delzer, J.A., Hawley, J.R., Romanyukha, A., Nemmers, S., Selwyn, R., Benevides, L.A., 2008. Long-term fade study of the DT-702 LiF: Mg, Cu, P TLD. Radiat. Prot. Dosimetry 131, 279e286. Eisenbud, M., Gesell, T., 1997. Environmental Radioactivity: From Natural, Industrial and Military Sources, fourth ed. Academic Press, New York. Gilvin, P.J., 2007. Comparison of time effects, Decision limit and residual signal in Harshaw LiF: Mg, Ti and LiF: Mg, Cu, P. Radiat. Prot. Dosimetry 125, 233e236. Horowitz, A., Horowitz, Y.S., 1992. Elimination of the residual signal in LiF: Cu, Mg, P. Radiat. Prot. Dosimetry 40, 265e269. Jones, L.A., Stokes, R.P., 2007. Pre-irradiation and post-irradiation fading of the Harshaw 8841 TLD in different environmental conditions. Radiat. Prot. Dosimetry 125, 241e246. Knoll, G.F., 2000. Radiation Detection and Measurement, third ed. John Wiley & Sons, Inc, New York. Moscovitch, M., 1999. Personnel dosimetry using LiF: Mg, Cu, P. Radiat. Prot. Dosimetry 85, 49e56. Moscovitch, M., St John, T.J., Cassata, J.R., Blake, P.K., Rotunda, J.E., Ramlo, M., Velbeck, K.J., Luo, L.Z., 2006. The application of LiF: Mg, Cu, P to large scale personnel dosimetry: Current status and future directions. Radiat. Prot. Dosimetry 119, 248e254. Ramlo, M., Moscovitch, M., Rotunda, J.E., 2007. Further studies in the reduction of residual in Harshaw TLD-100H (LiF: Mg, Cu, P). Radiat. Prot. Dosimetry 125, 217e219. Romanyukha, A., King, D., Benevides, L.A., 2008. Effect of background radiation on the lower limit of detection for extended dosemeter issue periods. Radiat. Prot. Dosimetry 131, 180e187. Systat, 2004a. SigmaPlot for Windows. Systat, 2004b. SigmaStat for Windows. Tang, K., et al., 2002. Influence of Readout Parameters on TL response, re-usability and residual signal in LiF: Mg, Cu, P. Radiat. Prot. Dosimetry 100 (1e4), 353e356. Traino, A.C., Perrone, F., Luperini, C., Tana, L., Lazzeri, M., d’Errico, F., 1998. Influence of background exposure on TLD minimum dose detection and determination limits. Radiat. Prot. Dosimetry 78, 257e262. Taylor, J.R., 1997. An Introduction to Error Analysis. University Science Books, Sausalito, California.
Contents lists available at ScienceDirect
Radiation Measurements journal homepage: www.elsevier.com/locate/radmeas
Long-term neutron dose fading study of LiF: Mg,Cu,P TLD J.A. Delzer a, A. Romanyukha b, *, L.A. Benevides b a b
Uniformed Services University of the Health Sciences, Bethesda, MD 20814, USA Naval Dosimetry Center, 8901 Wisconsin Avenue, Bethesda, MD 20889, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 5 October 2009 Received in revised form 5 March 2010 Accepted 10 November 2010
An extensive study of neutron dose fading in LiF:Mg,Cu,P thermoluminescent dosimeters (TLD) using commercially-available dosimeters and equipment for their processing was undertaken. During the 52 week study, three thousand TLDs were stored for various lengths of time before and after being exposed to a plutonium beryllium radiation source. The TLDs were subsequently processed and the resulting doses were compared to the reference exposure. Both a loss of signal and a loss of sensitivity were evaluated. The results of this study have shown that the commercially produced LiF:Mg,Cu,P TLD has no statistically significant change in sensitivity or change in signal with up to 52 weeks of pre-irradiation or post-irradiation time. The results of this study will provide the technical basis for increasing the exposure record accuracy, and extending the usable lifetime of dosimeters. Published by Elsevier Ltd.
Keywords: TLD Dosimetry Neutron Dose fading TLD-600H TLD-700H
1. Introduction Since 2002, U.S. Navy has used the dosimetric system based on LiF:Mg,Cu,P thermoluminescence dosimeters (TLD’s) and Harshaw model 8800 dose readers, developed and produced by Thermo Fisher Scientific (for more details, see Cassata et al., 2002; Delzer et al., 2008; Moscovitch, 1999; Moscovitch et al., 2006). The dosimeter consists of four LiF:Mg,Cu,P elements, mounted on an aluminum substrates which were placed in a plastic holder with the appropriate filtration. Three elements, TLD-700H (7LiF:Mg,Cu,P), are sensitive only to photon and beta irradiation, whereas the fourth element of TLD-600H (6LiF:Mg,Cu,P), is sensitive to photon, beta irradiation and thermal neutron resulting from albedo irradiation. The neutron dose measurement by TLD-600H is based on the following nuclear reaction. 6
Li þ 1 n/3 H þ 4 aþ
(1)
Thus the neutron dose reading is caused by energy deposition of H and 4aþ ions in TLD-600H which is also sensitive to photon and beta irradiation. The neutron dose is calculated as the by taking the difference of the element 4 and element 1 responses. From reaction (1), one can see that as result of each neutron exposure the amount of neutron-sensitive 6Li [(sensitive to neutron)] in element 4 3
* Corresponding author. Tel.: þ1 301 295 4149; fax: þ1 301 295 4165. E-mail address: [email protected] (A. Romanyukha). 1350-4487/$ e see front matter Published by Elsevier Ltd. doi:10.1016/j.radmeas.2010.11.012
steadily decreases. To make things more complex, the tritium generated by neutron exposure decays to 3He with emission of a beta particle (average energy of 5.7 keV) and antineutrino. Halflife of tritium is 12.33 years. Thus, besides the reduction of sensitivity (6Li reduction in element 4), neutron irradiation also results in self-irradiation of the element 4 and the generation of gaseous 3 He. Plus neutron dose determination as difference of the doses of element 1 and element 4 can be affected by possible photon dose fading in element 1. Clearly these facts make it important to study the effects of neutron pre-irradiation fade (sensitivity) and neutron post-irradiation (signal) fade. Very few studies on the neutron preand post-irradiation fading properties of this material have been performed (see Jones and Stokes, 2007; Gilvin, 2007). Most detailed previous study of neutron dose fading (Jones and Stokes, 2007) had established that the main dosimetric signal for this material was stable over a storage time of about 100 days and concluded that the pre-irradiation and post-irradiation fading of a commercially available LiF:Mg,Cu,P TLD irradiated in a mixed photon-neutron field at various storage temperatures were relatively small over a 24 week period. The maximum loss of sensitivity to neutron exposure and neutron signal did not exceed 8% at room temperature at the high storage temperature of 50 C the loss was found 18%. This study was performed with only 40 dosimeters at each of three storage temperature, e.g. 5 dosimeters per one time point, e.g. 0, 7, 14, 28, 56, 84 and 164 days. In difference from our results for photon dose fading in TLD-700H (Delzer et al., 2008; Jones and Stokes, 2007) found a significant increase in sensitivity after 60 days of storage at room temperature. Therefore it is important to clarify
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204 500
80000 70000
400 60000 50000
300
40000 200
30000
Temprerature, C
that observed effects statistically significant at bigger numbers of dosimeters and time points. Present paper represents the neutron portion of a two-phase study that evaluated the photon (Delzer et al., 2008) and neutron fading characteristics of the LiF:Mg,Cu,P TLD (Harshaw Model 8840/41). The methods used for this study were consistent with the previous published methods (Delzer et al., 2008). The main purpose was to determine if the LiF:Mg,Cu,P TLD has a statistically significant loss of sensitivity or loss of signal with either neutron pre-irradiation or neutron post-irradiation fade over a period of 52 weeks on 15 dosimeters for each time point.
TL response, arbitrary units
200
20000 100 10000
2. Materials and methods 0
The TLD, Harshaw Model 8840/41, consists of the TLD card and holder. The card contains four TeflonÒ encapsulated LiF:Mg,Cu,P elements. The holder contains tin, copper, Mylar and plastic filters that allow for discrimination of gamma radiation energy levels and determination of beta and neutron dose information (Horowitz and Horowitz, 1992). The TLD reader used for this study was the traditional Harshaw Model 8800 TLD Reader. A reader calibration factor (RCF) was determined each day, prior to processing any TLDs. Variations in measurements in the TLD reader are caused by environmental changes and are corrected by the RCF. Proper calibration of a reader and dosimeters used in this study is a critically important because of significant length of the study. Details of the used calibration procedure can be found in Delzer et al. (2008). All TLDs were annealed and processed using a time-temperature profile (TTP) that was designed to maximize the sensitivity and reusability of the DT-702. The TTP began with an initial preheat of the TLD to 165 C for 10 s to eliminate low temperature traps that are confounding and short lived compared to the main dosimetric peak (Budzanowski, 2002). This was followed by a 16.67 s acquisition time, where heat increased at a rate of 15 C/s to a maximum of 260 C. The TL response was determined by glow curve integration. Maximum temperatures were limited to 260 C to prevent damage to the TLD material from overheating (Cassata et al., 2002; Tang et al., 2002). A typical observed glow curve and TTP are shown on Fig. 1. No additional thermal treatments were used in the study. Three thousand TLDs were used in this study. The entire set of TLDs were annealed one day prior to the beginning of Day 0 of the study by performing two sequential 2-mrem (20mSv) anneals. This ensured that any residual signal was reduced to the lowest degree possible with the TTP used (Horowitz and Horowitz, 1992; Ramlo et al., 2007). The TLDs were sorted into several sets according to a pre-established irradiation schedule (Fig. 2). Once the sets of TLDs were established, irradiations were performed to emulate normal
0
50
100
400 TLDs stored with fade groups
Post Irradiation Fade 500 TLDs Irradiated to 4 mSv using PuBe Neutron Source Day 0
0 200
Channel Fig. 1. Solid line shows typical observed glow curve. Dash line shows applied timetemperature profile.
pre- and post-irradiation times that would be expected during normal handling. The first TLDs irradiated were the post-irradiation set, done on Day 0. This group was then used for the rest of the study for the Post-Irradiation fade group. The Pre-Irradiation TLD group was irradiated each week. Prior to processing, the Combined Pre/ Post Irradiation TLDs were irradiated to allow for mixed pre- and post-irradiation times. To account for any machine variance, a set of reference TLDs was utilized. This set consisted of a group of 15 TLDs and were processed on Day 0. Following Day 0, the reference set was irradiated along with the Pre-Irradiation TLDs and processed with them. Although these TLDs were annealed prior to irradiating, an additional annealing process was performed for consistency in study design to ensure there was no residual signal on the TLD. Neutron irradiation utilized a plutonium beryllium (PuBe) source and a carousel irradiation method developed specifically for this study. The developed irradiator was able to provide highlyinformed neutron exposure to 100 dosimeters simultaneously. The neutron source consisted of an encapsulated plutonium beryllium neutron source of 76.37 g of plutonium 239 (5 Ci) with a cylindrical shape measuring 3.30 cm (outside diameter) by 8.64 cm high and producing 2.3 106 neutrons per cm2-second per Curie. The source is surrounded by a virgin polyethylene shield to allow for thermalization of neutrons. Borated polyethylene is used to prevent escape of neutrons from the shielding. The delivered dose rate for the irradiation method used was 67 mSv per hour. Using a carousel irradiator, 100 TLDs were mounted on the carousel and evenly distributed keeping Element 4 of the TLD located on the bottom-
3000 TLDs annealed at start of study
Control TLDs
150
Pre Irradiation Fade 500 TLDs Irradiated to 4 mSv using PuBe Neutron Source Just prior to reading
Fig. 2. Establishment of TLD Groups used for fade study.
Mixed Pre and Post Fade 1600 TLDs Irradiated to 4 mSv using PuBe Neutron Source and stored for varying times
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204
201
Fig. 3. a) The top view of the carousel used to irradiate the TLDs. (b) A cut-away view shows the position of the PuBe source in relation to the TLDs cards. It is important to note that the carousel was placed in motion prior to inserting into source and remained rotating after the irradiation ensuring a homogeneous irradiation.
inner radius to be irradiated as indicated in Fig. 3. The TLDs were irradiated to 4 mSv. Rotation of the carousel was initiated prior to irradiation, and continued until the carousel was removed from the PuBe source. The TLDs were removed from the carousel assembly and prepared for processing. 3. Results and discussion Throughout the course of the study, the average of the group was determined, and then evaluated for outliers. The contribution of the control TLDs was subtracted and the data was normalized to the baseline data determined at the start of the study. It should also be noted that the propagation of error was carried through for the entire study. The average dose of each element for each set of TLDs was determined for each data point and the associated standard deviation of each set of data was calculated as the standard error. After processing each set of TLDs, the results of each element were evaluated for outliers to identify any statistically spurious data points using Chauvenet’s criterion (Taylor, 1997). This process was performed once for each set of TLDs (both background and fade groups) obtaining a final mean value and standard error for each data point. After evaluating each set of TLDs for outliers, the mean of the control TLDs was subtracted to give a background corrected reading for each element of the TLD. Standard error propagation methods were utilized as described by Knoll (2000). After correcting each data point for background, the data was normalized against the week 0 TLD results or baseline data, by taking a ratio between Week 0 and Week X. This provided a relative response of signal or sensitivity based on a group of TLDs with no associated pre-exposure or post-exposure fade. The evaluation of the resulting data focused primarily on the determination of fading over time, specifically whether a statistical difference existed in the change of signal and sensitivity for pre, post, and combination of fade for the entire 52 week study. In order to perform this evaluation, some assumptions were made on the specific dosimeter characteristics that included the statistical distribution and the expected variations for the TLD. The test for significant differences was set at a 95% confidence interval and performed among sample means (m) of the independent variable (known signal applied to TLD), and the dependent variable (time). The statistical single element variation in the dosimeter has been reported to be 2e4% (Moscovitch, 1999). In this study, a value of 3.5% was used for reproducibility based on the historical average observed by the Naval Dosimetry Center. The variation was
determined by repeated measurements of the same card with the same dose. Given that the distribution of dose is Gaussian (normal) in nature for the entire population of TLDs, it should be expected that 99% of the readings will fall within three standard deviations (3s) of the average. Therefore, a maximum 3s experimental error of about 10.5% would be expected. While this was not mathematically accounted for in this study, it can be assumed that differences of about 10.5% or less are well within the experimental error. One-way analysis of variance (ANOVA) tests were performed on the results of each element for all variations of storage time. For all ANOVA calculations, an alpha level of 0.05 was used. In the event that a statistical difference was detected by the ANOVA, a multiple comparison test using the Holm-Sidak method was performed (Systat, 2004a). The Holm-Sidak test is a non-parametric test that is more powerful than the standard ANOVA. The test attempts to identify the specific points that are statistically different compared to the control value (no storage time baseline response) (Systat, 2004a). When performing the test, the P values of all comparisons are computed and ordered from smallest to largest. Each P value is then compared to a critical level against a 95% significance level, the rank of the P value, and the total number of comparisons made. A P value less than the critical level (0.05) indicates there is a significant difference between the corresponding two groups. Throughout the study, control TLDs were used to account for background radiation, present from natural and manmade sources. Environmental radiation accounts for approximately 1 mSv per year of the total annual effective dose equivalent in the United States of approximately 3.6 mSv per year (Eisenbud and Gesell, 1997). In this study, the TLDs were stored in an area of much lower background resulting in a daily rate of 1.124 mSv/day or 0.409 mSv/year. Study TLDs were processed weekly and background radiation levels were subtracted from the study groups. In analyzing the background level, two analyses were performed; determination of a consistent daily background rate, and statistical difference between the different elements for the same groups of TLDs across the entire study period. In theory, the rate should remain constant over time and since the contribution to background is primarily from photons, all four elements should have the same background over time. To analyze these data points, a simple linear regression was used to evaluate the daily rate and a Pearson Product-Moment Correlation was used to compare the correlation of the first element with the other three (Systat, 2004b). The Pearson Correlation showed no statistical difference between all 4 elements when individual elements were compared against each other.
202
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204
Average Background Accumulation 450
Background Radiation Accumulation (µSv)
400
350
300
250
200
150
100
y = 7.3077x + 6.2041 R2 = 0.9982
50
0 0
10
20
30
40
50
Week Fig. 4. The graph shows the average of all elements background radiation accumulation. Linear regression is shown as a solid line through the data points. The error bars indicate the propagated errors for the mathematical manipulation (1s). Equation of the regression analysis and R2 value for linear regression fit are listed.
The linear regression for each element resulted in an average R2 value of 0.999. Fig. 4 shows the regression plot, associated equation and R2 value. For the study, the primary focus was to determine the effects of fading due to neutron irradiation, based on results from Element 4. A One-way ANOVA test performed on the Pre-Irradiation TLD group for the combined photon and neutron contributions indicated that there was no significant difference between the responses of the elements for the entire 52 week period. (P ¼ 0.707). The maximum increase was measured at 5 weeks at 5%. The TLD response for sensitivity appeared to remain consistent for entire study period, with a maximum decreases measured at only 6%. To analyze the pure neutron component of pre-irradiation fade, the photon component from Element 1 must be subtracted from Element 4. In order to ensure the geometry was consistent, a study was performed to account for the differences in irradiation position. This was performed by rotating the TLD card through all the positions of the carousel and determining the ratio between position 1 and position 4 for pure photon. In order to apply a correction factor, it must be determined that there is truly a significant difference between the two values due to geometry. This was accomplished by performing a t-test for Element 1 geometry positions (Systat, 2004b). The result of this test identified a significant difference between the two positions (P 0.001). Therefore, when calculating a ratio, a difference of 8% due to the geometry differences that exist between Element 1 and Element 4. After the correction for Element 1 was determined, it was subtracted from the value of Element 4. This methodology was also used by Jones and Stokes, 2007 in determining the pure neutron effects. After performing the neutron calculation, the effects of fade from pure neutron were analyzed. A One-way ANOVA test performed on the Pre-Irradiation TLD group for the neutron contributions indicated that there was no statistically significant difference between the responses of the elements for the entire 52 week period. (P ¼ 0.754). The maximum increase was measured at 5 weeks at 5%. The TLD response appeared to remain consistent for entire study period, with a maximum decreases measured at only 6%. Changes in sensitivity for Element 4 for only neutron are displayed in Fig. 5. An interesting finding in analyzing the data is that
there is little difference between pre-irradiation fade for photon and neutron irradiation. This is not surprising, since the contribution from the photon component was only 4% of the total signal. For that reason, the data provides a minimal insight to the differences of photon and neutron fading characteristics. Element 4 was analyzed for Post-Irradiation Fade, which is considered a change of signal over time. A One-way ANOVA test performed on the Post-Irradiation TLD group for the combined photon and neutron contributions indicated that there was no significant difference between the responses of the elements for the entire 52 week period (P ¼ 0.312). The maximum increase was measured at 5 weeks at 5%. The TLD response appeared to remain consistent for the entire study, with maximum decreases measured at only 2%. After performing the same correction for neutron contribution as used for Pre-Irradiation Fade, the effects of fade from pure neutron were analyzed. A One-way ANOVA test performed on the Post-Irradiation TLD group for the neutron contributions indicated that there was no significant difference between the responses of the elements for the entire 52 week period. (P ¼ 0.766). The
Fig. 5. The plot displays the trend for Neutron pre fade normalized response over a 52 week period. The error bars indicate the propagated errors for the mathematical manipulation (1s).
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204
Fig. 6. The plot displays the trend for Neutron post fade normalized response over a 52 week period. The error bars indicate the propagated errors for the mathematical manipulation (1s).
Fig. 7. The plot displays the trend for equal time between pre- and post- neutron fade normalized response over a 52 week period. The error bars indicate the propagated errors for the mathematical manipulation (1s).
maximum increase was measured at 39 weeks at 6%. The TLD response appeared to remain consistent for entire study period, with a maximum decreases measured at only 3%. Changes in signal for Element 4 for only neutron are displayed in Fig. 6.
203
Fig. 8. The plot displays the trend for the combined pure pre- and post- neutron fade normalized response over a 52 week period. The error bars indicate the propagated errors for the mathematical manipulation (1s).
An analysis was performed on Combined Fade results. Several different combinations are presented with varying results in the data. First, an analysis was done on the combined pure Pre and Post Fade data, then an analysis was performed on equal time between Pre and Post data, and finally an analysis was performed on all of the data. Analyzing the pure Pre and Post results for combined neutron fade with one-way ANOVA resulted in no statistically significant difference (P ¼ 0.860). The results can be seen in Fig. 7. The second analysis performed for the combined fade was determining the effects of equal time between pre and post-exposure time. This group consists of all TLDs that had an equal amount of time between the anneal-expose-read periods. One-way ANOVA initially detected a significant difference (P ¼ 0.008), but HolmSidak testing determined no point to be statistically significant (P ¼ 0.0114). The neutron equal pre and post fade results can be seen in Fig. 8. Finally, a statistical analysis was performed on the entire population of pre and post-exposure fade data. One-way ANOVA initially detected a significant difference (P ¼ 0.006), but the nonparametric Holm-Sidak testing determined no point to be statistically significant (P ¼ 0.0124). This result is quite remarkable considering there are 237 data points, accounting for 2474
Fig. 9. Time dependence of combined pre- and post- neutron dose fade normalized response.
204
J.A. Delzer et al. / Radiation Measurements 46 (2011) 199e204
TLDs for all the fade groups, over the entire 52 week period (see Fig. 9). 4. Conclusions The LiF:Mg,Cu,P thermoluminescence dosimeters have been shown to have no significant change in sensitivity or signal response with up to 52 weeks of neutron pre-exposure, postexposure, or combination of pre-exposure and post-exposure time. We have not observed previously reported in Jones and Stokes (2007); Gilvin, 2007 variations in both sensitivity (at 60 days storage) and signal for neutron and photon exposures. The differences in conclusions between this study and previous ones (Jones and Stokes, 2007; Gilvin, 2007) can be attributed to the greater statistical power that is obtained with the use of so many dosimeters. Currently the TLDs are normally issued by U.S. Navy for eightweek periods, with up to 12-week maximum for special circumstances. The data presented in this study suggest that extending beyond these limits may be warranted. With this increase, flexibility with international and domestic shipping procedures, as well as reduced workload requirements for dosimetry processing could be realized. When evaluating issue periods, environmental conditions should coincide with the ones established by this study design. This extension of issue periods must be concurrently balanced with the effect of lower limit of detection (LLD), since an increase in LLD may be seen with longer issue periods (see for detail Traino et al., 1998; Romanyukha et al., 2008). Acknowledgements The study was funded through U.S. Department of Defense operational and maintenance budget. JAD would like to give special thanks to the following people for their contributions provided during this research: Dr. Tomoko Hooper, Uniformed Services University of the Health Sciences, for her honest feedback and support throughout. CDR Russ Lawry, Uniformed Services University of the Health Sciences, for his mentorship and questioning attitude throughout the study,
Mr. Dave King, Naval Dosimetry Center, for his contributions in the study design and technical support. The views expressed in this paper are those of the authors and do not reflect the official policy or position of the Navy and Marine Corps Public Health Center, Navy Bureau of Medicine and Surgery, Department of the Navy, Department of Defense or the U.S. Government. References Budzanowski, M., 2002. The Influence of post-exposure heating on the Stability of MCP-N (LiF:Mg, Cu, P) TL Detectors. Radiat. Prot. Dosimetry 101 (1e4), 257e260. Cassata, J.R., Moscovitch, M., Rotunda, J.E., Velbeck, K.J., 2002. A New Paradigm in Personal dosimetry using LiF: Mg, Cu, P. Radiat. Prot. Dosimetry 101, 27e42. Delzer, J.A., Hawley, J.R., Romanyukha, A., Nemmers, S., Selwyn, R., Benevides, L.A., 2008. Long-term fade study of the DT-702 LiF: Mg, Cu, P TLD. Radiat. Prot. Dosimetry 131, 279e286. Eisenbud, M., Gesell, T., 1997. Environmental Radioactivity: From Natural, Industrial and Military Sources, fourth ed. Academic Press, New York. Gilvin, P.J., 2007. Comparison of time effects, Decision limit and residual signal in Harshaw LiF: Mg, Ti and LiF: Mg, Cu, P. Radiat. Prot. Dosimetry 125, 233e236. Horowitz, A., Horowitz, Y.S., 1992. Elimination of the residual signal in LiF: Cu, Mg, P. Radiat. Prot. Dosimetry 40, 265e269. Jones, L.A., Stokes, R.P., 2007. Pre-irradiation and post-irradiation fading of the Harshaw 8841 TLD in different environmental conditions. Radiat. Prot. Dosimetry 125, 241e246. Knoll, G.F., 2000. Radiation Detection and Measurement, third ed. John Wiley & Sons, Inc, New York. Moscovitch, M., 1999. Personnel dosimetry using LiF: Mg, Cu, P. Radiat. Prot. Dosimetry 85, 49e56. Moscovitch, M., St John, T.J., Cassata, J.R., Blake, P.K., Rotunda, J.E., Ramlo, M., Velbeck, K.J., Luo, L.Z., 2006. The application of LiF: Mg, Cu, P to large scale personnel dosimetry: Current status and future directions. Radiat. Prot. Dosimetry 119, 248e254. Ramlo, M., Moscovitch, M., Rotunda, J.E., 2007. Further studies in the reduction of residual in Harshaw TLD-100H (LiF: Mg, Cu, P). Radiat. Prot. Dosimetry 125, 217e219. Romanyukha, A., King, D., Benevides, L.A., 2008. Effect of background radiation on the lower limit of detection for extended dosemeter issue periods. Radiat. Prot. Dosimetry 131, 180e187. Systat, 2004a. SigmaPlot for Windows. Systat, 2004b. SigmaStat for Windows. Tang, K., et al., 2002. Influence of Readout Parameters on TL response, re-usability and residual signal in LiF: Mg, Cu, P. Radiat. Prot. Dosimetry 100 (1e4), 353e356. Traino, A.C., Perrone, F., Luperini, C., Tana, L., Lazzeri, M., d’Errico, F., 1998. Influence of background exposure on TLD minimum dose detection and determination limits. Radiat. Prot. Dosimetry 78, 257e262. Taylor, J.R., 1997. An Introduction to Error Analysis. University Science Books, Sausalito, California.