- Email: [email protected]
Equivalent dose of quartz originating from ceramics obtained by OSL SAR method – Tests of protocol parameters
Radiation Measurements xxx (2015) 1e6
Contents lists available at ScienceDirect
Radiation Measurements journal homepage: www.elsevier.com/locate/radmeas
Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Tests of protocol parameters ska N. Kijek*, A. Chruscin Institute of Physics, Faculty of Physics, Astronomy and Informatics, Nicolaus Copernicus University, Grudziadzka 5, Torun, Poland
h i g h l i g h t s 160 C and 200 C TL traps in quartz from brick influence the OSL process at 125 C. The De from SAR protocol for brick samples depends on OSL measurement temperature. OSL component analysis for different measurement temperatures is presented. Lower temperature of OSL readout in SAR protocol for quartz from bricks is suggested.
a r t i c l e i n f o
a b s t r a c t
Article history: Available online xxx
In the standard OSL dating procedure, the OSL signal is measured at 125 C. For quartz samples from bricks the presence of traps responsible for the 160 C and 200 C TL peaks has an influence on the OSL process, just like the one for the traps responsible for the 110 C TL peak. In order to examine the significance of this disturbance for dating results, a series of measurements of De was carried out for various temperatures of the OSL readout in SAR protocol. For most samples, significant fluctuations of De values were obtained, especially at temperatures above 90 C. Also the analysis of OSL components carried out for curves obtained at different measurement temperatures shows that the TL traps active up to 200 C make the OSL process more complex. The measurement at 125 C does not assure the exclusion of shallow traps from the OSL process. A reasonable solution of this problem may be the choice of temperatures below the range of dominant TL peaks, in the SAR protocol for OSL measurements. © 2015 Published by Elsevier Ltd.
Keywords: Optically stimulated luminescence Optical cross-section Quartz Cermics dating
1. Introduction Ceramics e pottery and bricks e was one of the first objects to undergo thermoluminescence (TL) dating over fifty years ago (Kennedy and Knopf, 1960; Goedicke et al., 1981). The investigations of this material have also made an important contribution to the development of retrospective dosimetry. Recently, the OSL method is more preferably applied in the ceramics dating than the TL method (Guibert et al., 2009; Blain et al., 2010; Bailiff, 2007). The SAR protocol used in dating (Wintle and Murray, 2006; Wintle, 2008), was developed for geological sediments. The parameters of OSL measurements, like the temperature, the optimal stimulation time and the manner of bleaching the signal after OSL regenerated measurements were established. Although in both TL and OSL methods, quartz is used, it is commonly known that the
* Corresponding author. E-mail address: natalia@fizyka.umk.pl (N. Kijek).
luminescent properties of quartz annealed to high temperatures during pottery production differ from the properties of quartz ska et al., 1996; Petrov and originating from sediments (Chruscin Bailiff, 1997). The characteristic effect observed in annealed quartz is the growth in intensity of the TL peak at about 210 C, which is significantly less stable then the signal used in dating ska et al., 1996; Bailiff and Petrov, 1999). Therefore some (Chruscin caution should be maintained in using the same protocol for dating sediments and ceramics, especially since for sedimentary quartz some fluctuations were observed in the value of equivalent dose (De) for measurement temperatures at about 125 C (Kijek et al., 2013). In the standard dating procedure, the OSL signal is read out at 125 C. The presence of traps responsible for the 210 C peak has an influence on the OSL process, just as the traps responsible for the 110 C peak do. Hence the temperature of OSL measurement used in the SAR protocol should be selected with special care. In order to examine its influence on dating results, a series of measurements of De were carried out for various temperatures TOSL
http://dx.doi.org/10.1016/j.radmeas.2015.02.003 1350-4487/© 2015 Published by Elsevier Ltd.
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
2
during readout of the OSL signal in the SAR protocol, namely for: 25, 55, 90, 125, 160 and 195 C. For most samples significant fluctuations of De values were obtained as described below. TL measurements may provide a useful insight into carrier transfer processes during the steps of a SAR protocol and may explain the origin of De variations. TL was measured before each of seven initial steps of SAR. Changes of the TL curve shape in subsequent steps were observed and compared for all examined samples. Two groups of samples were distinguished which behave in a different way in the test of De dependence on temperature. 2. Experimental Quartz grains have been separated from 11 ceramics samples labeled: JK0, JK1, JK2, JK3, JK4, JK5, JAK2, JAK3, JAK4, JAK5, JAK6 in the way common for optically stimulated luminescence (OSL) dating. Mineral grains about 100e200 mm in diameter have been obtained for single aliquot measurements. Aliquots of 4 mm diameter have been prepared using Silkospray silicone oil to mount grains. Multiple grain (MG) SAR measurements as well as TL measurements have been carried out using a Risø TL/OSL reader. A blue LED stimulation module (470 nm, 80 mW cm2) has been used as an optical stimulation source for the MG method. For artificial irradiation, a 90Sr/90Y b source giving a dose rate of about 10 mGy s1 has been utilized. Another light source with a wavelength of 520 nm, similar to the wavelength of the laser used in the Risø Single Grain Attachment, has been also used for OSL measurements; it was fitted to a Risø TL/ OSL System TL-DA-12 equipped with an EMI 9235QA PMT and 2 mm Schott UG11-IRB filter. A strong Xenon lamp with a monochromator (180e1000 nm) and lightquide with a special adapter has been applied instead of a halogen lamp normally used for stimulation in TL-DA-12 readers. The photon flux used for stimulation has been 8.8 1016 cm2 s1 and the bandwidth of the stimulation was 26 nm. During irradiation a 90Sr/90Y source giving dose rate of 40 mGy/s was used. 3. Results 3.1. De dependency on the temperature of OSL measurement In order to determine De the single aliquot regeneration dose protocol (SAR) has been applied, which is typically used for age calculation. A preheat test has been performed before the actual study for each sample (from 180 C to 240 C, every 20 C) in order to find out the proper temperature of preheat for the standard SAR protocol. The whole SAR protocol has been repeated for six different temperatures TOSL of OSL readout: 25 C, 55 C, 90 C, 125 C, 160 C, 195 C. Sixteen aliquots of each sample have been used for estimating an individual De value for each temperature. The above-mentioned series of SAR protocol measurements have been applied for each of the 11 samples. Besides the observation of the dependence of De on the temperature, the repeatability of the results (the recycling test) and the quality of bleaching the OSL signal after a regeneration dose (the recuperation test) have been monitored. For each sample a dose recovery test was performed. In this case a known laboratory dose of the order of De has been reconstructed by applying the SAR protocol. Fig. 1 presents the results of De estimation for different temperatures of OSL measurement in SAR protocol. The shown values have been normalized to De obtained at 25 C. Fig. 2 shows similar results for the dose recovery test and Fig. 3 provides recycling test outcomes for each sample. The presented De values are obtained after discarding aliquots, for which De values differ significantly from the average
Fig. 1. The results of De estimation for different temperatures of the OSL measurement. Each value is normalized to De obtained at 25 C. For clarity of the figure the subsequent plots have been shifted upwards by the value 0.2. The grid lines are marked at values (0.9 þ i 0.2) and (1.0 þ i 0.2), where i ¼ 0, 1, …, 10 for each plot.
calculated for all 16 aliquots or for which the recycling ratio was outside the range 0.9e1.1. One can clearly see that higher temperatures of OSL measurement lead to a strong disturbance of De values for most samples. Although the character of these changes is sample-dependant, one can observe that the most intensive changes appear in the temperature region from 90 to 200 C, which is the characteristic temperature of intense TL peaks in quartz. There is, however, a group of samples, JK2, JK3, JAK2, JAK4 and JAK6, for which De variations are small and do not exceed 10%. Among them, only for four samples, JK3, JAK2, JAK4 and JAK6, have positive results been obtained also in the recovery and recycling tests. For the rest of the samples the De variations with measurement
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
Fig. 2. The results of recovery dose estimation for different temperatures of the OSL measurement. Each value is normalized to De obtained at 25 C. For clarity of the figure the subsequent plots have been shifted upwards by the value 0.2. The grid lines are marked at values (0.9 þ i 0.2) and (1.0 þ i 0.2), where i ¼ 0, 1, …, 10 for each plot.
temperature are larger than 10% and mostly bigger for higher temperatures. Some results of the recovery and recycling test for these samples, are nevertheless acceptable, as for example in the case of sample JK5 (both tests acceptable) or JK1, for which the recovery test has given positive results. Very good results for the recycling test have been obtained also for sample JAK5. 3.2. TL measurements One way to look inside OSL kinetics during SAR measurements is to observe TL curve changes after subsequent steps of this protocol.
3
For each investigated sample, TL curves were measured (with a heating rate 5 K s1) at each stage of the protocol, so a set of 7 TL curves was created: natural TL, natural TL after preheat (the temperature established by preheat test for OSL), natural TL after OSL measurement (40 s at 125 C), laboratory TL (after test dose), laboratory TL after preheat, laboratory TL after OSL measurement, laboratory TL after OSL bleaching at 280 C. Each TL curve has been measured for 3 fresh aliquots that individually have undergone adequate number of SAR steps. Fig. 4 illustrates the key findings from the analysis of the outcome of TL measurements. The natural as well as laboratory TL curves differ in the intensities of individual TL peaks (Fig. 4a, b). A peak about 200 C dominates in the natural curves. The peak at 110 C is also present before preheat. In the laboratory excited curves the peak at 110 C dominates and apart from the peak at 200 C there appears a peak at 160 C of a comparative intensity. Here, it should be stressed that mutual relations between the intensities of individual peaks which are seen in TL curves do not reflect the real relation between the concentrations of traps responsible for those peaks. The very low intensity of high-temperature peaks may be caused by thermal quenching (Wintle, 1975). As it was demonstrated recently, taking into account this phenomenon and correcting the TL curve reveals that the concentration of high-temperature traps can be comparable to the concentration of the traps responsible for the 110 C TL peak (Subedi et al., 2012). The analysis of TL curves distinguishes two groups of the investigated samples that correlate with the groups identified on the basis of De variability with the temperature of OSL measurement. In Fig. 4a natural curves for the representatives of both sample groups (JK3 and JAK5) are compared. In the case of samples JK2, JK3, JAK2, JAK4, and JAK6, the 200 C peak dominates, whereas the 110 C peak is hardly observed. In the remaining samples the intensity of the latter peak is similar to the intensity of the 200 C peak. As mentioned previously, in the laboratory excited TL curves (Fig. 4b), the 110 C peak prevails, but in the curves of the samples listed above high-temperature peaks can still be clearly seen, although they are about 6 times weaker. This is not the case for other samples. The intensity of the 110 C peak is about 50 times greater than the peaks at 160 and 200 C. Another TL feature that differentiates the two sets of samples is presented in Fig. 5. It is connected with the presence of the peak just above 200 C in the natural as well as laboratory TL curves measured after preheat and OSL stimulation. In the curves of samples JK0, JK2, JK3, JAK2, and JAK4 this peak is always noticeable, even after a preheat at 240 C (Fig. 5a), but in the other samples it is not present (Fig. 5b) after preheat at 200 C. The next difference is that the decrease in the intensity of the TL curve after the OSL stimulation is evident in the curves for the first sample set but for the remaining samples the decrease in the TL peak intensity is hard to notice or, as it is illustrated for sample JAK5, one can observe the peak increase after optical stimulation. 3.3. OSL component study Observing the fluctuations of De it is worth testing the changes of OSL curve composition with the OSL measurement temperature. In order to find the OSL signal components a procedure of fitting the sum of first order OSL curves to the experimental OSL curve has been applied. The non-negative least squares procedure available in MATLAB has been used. In all outcomes the fitting of first order components to the OSL curves was satisfactory (FOM (figure of merit) for fitting results are always less than 2%). An example of the outcomes of the fitting procedure (Fig. 6.) shows the component OSL curves and their sum, together with the OCS (optical crosssection) spectra, i.e. intensities obtained for individual OCS values
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
4
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
Fig. 3. Results of the recycling test for SAR protocol carried out with various temperatures of OSL measurement. For each temperature an average of the results for 16 aliquots is given.
Fig. 4. Natural (a) and b) laboratory excited (b) TL curves for samples JK3, JAK5 and JK0. The heating rate used in TL measurements was 5 K s1, the dose applied for TL excitation e 0.75 Gy. In part (a) a laboratory excited TL curve for sample JK3 is added in order to facilitate the shape comparison for the both kinds of curves.
from the range 1024e1014 cm2 presented in the form of a column plot. Fig. 7 presents the results of decomposition of the OSL curves measured at different temperatures (25, 55, 90, 125, 160 and 195 C) into first-order components. Sample JAK5 was chosen as a representative of the samples that manifest significant variations of De. The OSL curve analysis was carried out for naturally and laboratory (beta dose of about 1 Gy) irradiated samples. The experiment for an individual temperature was repeated for three aliquots. The results were similar for natural and laboratory curves; hence they are presented only for the first case. Five components were found for all the tested temperatures. The most conspicuous observation concerns the behavior of the slow component. Its intensity fluctuates significantly at temperatures above 90 C. The intensities of all other components change their mutual relation in this temperature range, but the changes are not as radical as for the slow component. Simultaneously, constant decrease of the intensities of all components except for the slow one starts already above 60 C. This effect can be explained by the thermal quenching of luminescence. A similar series of experiments has been made for stimulation with light of 520 nm wavelength (bandwidth 26 nm). The outcomes
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
5
Fig. 5. TL curves (rate 5 K s1) obtained after the subsequent steps of SAR protocol: natural signal after preheat (black dash line) and after the OSL measurement (black solid line); laboratory TL after the preheat (light gray dash line) and after OSL measurement (light gray solid line). Examples are given for two samples e JK3 (a) and JAK5 (b) that are the good representatives of two sample sets that can be distinguished because of their TL features.
for natural OSL are shown in Fig. 8. For this lower stimulation energy the OCS spectra are less complex than for the stimulation with light at 470 nm. Three components are active in the whole range of temperatures. The OCS values for 520 nm stimulation are consistently lower than those for 470 nm with dependence of the OCS on the stimulation energy. The three components observed for 520 nm stimulation are probably the two fast and medium 1 components observed during the stimulation by blue diodes. Their intensities decrease monotonically in a similar way as was observed for 470 nm stimulation. The slow component, so intensive and extraordinary for 470 nm, is not present or rather sinks in the background. 4. Discussion The temperature of the OSL measurement in the SAR protocol applied for sediment dating was determined on the basis of investigations carried out for quartz grains from sediment (Murray
Fig. 6. The results of fitting the sum of the first order curves (crosses) to the experimental OSL curve (solid black line) and OSL components for a chosen aliquot. The bottom part e the difference between the curve obtained from fitting (sum of the OSL component) and the experimental OSL curve. Inset: the OCS spectra estimated by fitting procedure for all 3 aliquots measured in the same conditions. Spectra for the individual aliquots are presented in different colors.
Fig. 7. Results of OSL component analysis by fitting the sum of firsteorder OSL curves to an experimental curve. The dependence of the outcomes of fitting: the OCS values (a) and the initial concentration of electrons in traps (b) on the temperature of OSL measurement for the stimulation light of the wavelength of 470 nm (blue LEDs used in MG SAR). Each point on the graphs represents the average of 3 aliquots.
and Wintle, 1998). Reading out the OSL signal at a temperature about 125 C should keep the 110 C TL trap empty, which is responsible for slowing down the OSL process. In quartz from bricks that were annealed for a long time at high temperature during the brick production process, there are also traps that give TL peaks at about 160 and 200 C. These traps, contrary to what might be thought on the basis of the shape of the TL curve, have concentrations comparable to the concentration of the 110 C TL trap. As mentioned earlier, the low intensities of the peaks next to the 110 C peak are a consequence of the strong thermal quenching effect in quartz. The TL curves presented here for samples JK3 and JAK5 (Fig. 4) allow one to suspect that the concentration of 160 C TL traps can be even higher than the concentration of the 110 C TL traps. Therefore, the idea to avoid the slowing down of OSL process by shallower traps by keeping them empty should be also addressed to the 160 C and 200 C traps. This would require much higher temperature of OSL measurement in the SAR protocol. As the present study shows, this is not recommended because of the high fluctuations of the De values obtained in such readout conditions observed for many samples. Another disadvantage of the high OSL readout temperature is the impact of thermal quenching. It is, however, hard to attribute the poor results of De estimation for higher temperatures to this effect alone. The TL experiments allowed one to divide the investigated samples into two groups. The samples from the first group (JK2, JK3, JAK2, JAK4 and JAK6) are characterized by a high intensity of the 200 C TL peak, comparable to the intensity of the 160 C TL peak or higher (Fig. 4b, sample JK3), the clear optical bleaching of the TL signal in the range 300e400 C during the SAR protocol (Fig. 5a) and the presence of the 200 C TL peak in the TL signal even after the preheat at 240 C. This group demonstrates weak De fluctuations (below 10% of the value obtained at 25 C) and mostly for the
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
6
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
because of the 200 C TL traps) that may lead to significant fluctuations of De values with temperature and poor outcomes of the recycling and recovery tests. Therefore, one should consider changing the OSL measurement temperature in the SAR protocol to a significantly lower one, below the range of dominant TL peaks. In the range RT e 60 C the De fluctuations are less frequent for most samples. Simultaneously the recovery test for this temperature range gave good results for most of the samples. The investigation of the OSL components for 520 nm stimulation shows that this lower stimulation runs more stable (Fig. 8). Only three OSL components are active and the OCSs increase with temperature according to their normal behavior. Applying the 520 nm stimulation should be more intensively tested in application for OSL dating of ceramics. 5. Conclusions In quartz from ceramics the De value can strongly depend on the temperature of OSL measurement. The most significant fluctuations appear at temperatures above 90 C and are linked to poor results in the recycling test. The above effects are not so clearly evident in dose recovery experiments. In quartz from ceramics, beyond the 110 C TL traps there are also 160 and 200 C TL traps having very high concentrations, hence the 125 C temperature for OSL measurement in the SAR protocol does not assure the exclusion of the shallow traps from the OSL process. A reasonable solution to this problem may be applying a temperature below the range of the dominant TL peaks, in the SAR protocol for OSL measurements. Acknowledgments Fig. 8. Results of OSL component analysis by fitting the sum of firsteorder OSL curves to an experimental curve. The dependence of the outcomes of fitting: the OCS values (a) and the initial concentration of electrons in traps (b) on the temperature of OSL measurement for the stimulation light of the wavelength of 520 nm. Each point of the graphs represents the average from the results for 3 aliquots.
higher temperatures. The rest of samples have a very strong 160 C TL peak in comparison to the 200 C TL peak, and their TL signal in the range 300e400 C is not bleached during the optical stimulation in the SAR protocol (Fig. 5a). Presumably, traps deeper than the traps responsible for the TL signal in the range 300e400 C are the source of the OSL signal. It is, however, not clear why in the samples having lower concentration of the 160 C TL traps (JK2, JK3, JAK2, JAK4 and JAK6) the bleaching of TL in the range 300e400 C can be observed. The shallow TL traps have an impact on the individual components of the OSL signal (Fig. 7). Normally, the OCS values should increase with temperature. The activity of shallower traps disturbs this normal behavior in the range above 60 C. The OCSs fluctuate. It is most evident for the slow component. Intensities of all components decrease because of thermal quenching. The relation of the intensities of fast and slow components may be the most important for the quality of De estimation. The decrease of slow component intensity with temperature could be treated as a positive effect, however at the same temperature the ratio of slow component's OCS value to fast component's OCS value increases, so the part of the OSL related to the slow component in the total OSL signal stays as it was at lower temperatures. Summing up, the measurement of OSL at 125 C in the course of the SAR protocol applied to quartz from bricks seems to have no advantages such as increasing the total luminescence intensity or improving the ratio of the part of the OSL originated from the fast component to the slow component's part. Additionally the OSL kinetics in this temperature region is very complex because of the high concentration of 110 C and 160 C TL traps (and may be also
This work has been done thanks to the help of the Centre for Modern Interdisciplinary Technologies, Nicolaus Copernicus University in Torun, ul. Wilenska 4, 87-100 Torun, Poland (e-mail: [email protected]) and the grant No. 1765-F obtained from Nicolaus Copernicus University. References Bailiff, I.K., Petrov, S.A., 1999. The use of the 210C TL peak in quartz for retrospective dosimetry. Radiat. Prot. Dos. 84, 551e554. Bailiff, I.K., 2007. Methodological developments in the luminescence dating of brick from English late-medieval and post-medieval buildings. Archaeometry 49, 827e851. , M., 2010. An intercomparison Blain, S., Bailiff, I.K., Guibert, P., Bouvier, A., Bayle study of luminescence dating protocols and techniques applied to medieval brick samples from Normandy (France). Quat. Geochronol. 5, 311e316. ska, A., Oczkowski, H.L., Przegie˛ tka, K., 1996. Trap spectra of annealed Chruscin quartz. Acta Phys. Pol. A 89, 555e568. Goedicke, C., Slusallek, K., Kubelik, M., 1981. Thermoluminescence dating in architectural history: Venetian villas. J. Soc. Archit. Hist. 40 (3), 203e217. Guibert, P., Bailiff, I.K., Blain, S., Gueli, A.M., Martini, M., Sibilia, E., Stella, G., Troja, S.O., 2009. Luminescence dating of architectural ceramics from an early medieval abbey: the St Philbert Intercomparison (Loire Atlantique, France). Radiat. Meas. 44 (5e6), 488e493. Kennedy, G.C., Knopf, L., 1960. Dating by thermoluminescence. Archaeology 13, 147e148. ska, A., Przegietka, K.R., 2013. On the dependence of equivalent Kijek, N., Chruscin dose on temperature of OSL measurements for sediment quartz grains and its implication in dating practice. Radiat. Meas. 56, 252e256. Murray, A.S., Wintle, A.G., 1998. Factors controlling the shape of the OSL decay curve in quartz. Radiat. Meas. 29, 65e79. Petrov, S.A., Bailiff, I.K., 1997. Determination of trap depths associated with TL peaks in synthetic quartz (350e550 K). Radiat. Meas. 27, 185e191. Subedi, B., Polymeris, G.S., Tsirliganis, N.C., Pagonis, V., Kitis, G., 2012. Reconstruction of thermally quenched glow curves in quartz. Radiat. Meas. 47, 250e257 (of the Royal A). Wintle, A.G., 2008. Fifty years of luminescence dating. Archaeometry 50, 276e312. Wintle, A.G., 1975. Thermal quenching of thermoluminescence in quartz. Geophys. J. R. Astronomical Soc. 41, 107e113. Wintle, A.G., Murray, A.S., 2006. A review of quartz optically stimulated luminescence characteristics and their relevance in single aliquot regeneration dating protocols. Radiat. Meas. 41, 369e391.
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
Contents lists available at ScienceDirect
Radiation Measurements journal homepage: www.elsevier.com/locate/radmeas
Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Tests of protocol parameters ska N. Kijek*, A. Chruscin Institute of Physics, Faculty of Physics, Astronomy and Informatics, Nicolaus Copernicus University, Grudziadzka 5, Torun, Poland
h i g h l i g h t s 160 C and 200 C TL traps in quartz from brick influence the OSL process at 125 C. The De from SAR protocol for brick samples depends on OSL measurement temperature. OSL component analysis for different measurement temperatures is presented. Lower temperature of OSL readout in SAR protocol for quartz from bricks is suggested.
a r t i c l e i n f o
a b s t r a c t
Article history: Available online xxx
In the standard OSL dating procedure, the OSL signal is measured at 125 C. For quartz samples from bricks the presence of traps responsible for the 160 C and 200 C TL peaks has an influence on the OSL process, just like the one for the traps responsible for the 110 C TL peak. In order to examine the significance of this disturbance for dating results, a series of measurements of De was carried out for various temperatures of the OSL readout in SAR protocol. For most samples, significant fluctuations of De values were obtained, especially at temperatures above 90 C. Also the analysis of OSL components carried out for curves obtained at different measurement temperatures shows that the TL traps active up to 200 C make the OSL process more complex. The measurement at 125 C does not assure the exclusion of shallow traps from the OSL process. A reasonable solution of this problem may be the choice of temperatures below the range of dominant TL peaks, in the SAR protocol for OSL measurements. © 2015 Published by Elsevier Ltd.
Keywords: Optically stimulated luminescence Optical cross-section Quartz Cermics dating
1. Introduction Ceramics e pottery and bricks e was one of the first objects to undergo thermoluminescence (TL) dating over fifty years ago (Kennedy and Knopf, 1960; Goedicke et al., 1981). The investigations of this material have also made an important contribution to the development of retrospective dosimetry. Recently, the OSL method is more preferably applied in the ceramics dating than the TL method (Guibert et al., 2009; Blain et al., 2010; Bailiff, 2007). The SAR protocol used in dating (Wintle and Murray, 2006; Wintle, 2008), was developed for geological sediments. The parameters of OSL measurements, like the temperature, the optimal stimulation time and the manner of bleaching the signal after OSL regenerated measurements were established. Although in both TL and OSL methods, quartz is used, it is commonly known that the
* Corresponding author. E-mail address: natalia@fizyka.umk.pl (N. Kijek).
luminescent properties of quartz annealed to high temperatures during pottery production differ from the properties of quartz ska et al., 1996; Petrov and originating from sediments (Chruscin Bailiff, 1997). The characteristic effect observed in annealed quartz is the growth in intensity of the TL peak at about 210 C, which is significantly less stable then the signal used in dating ska et al., 1996; Bailiff and Petrov, 1999). Therefore some (Chruscin caution should be maintained in using the same protocol for dating sediments and ceramics, especially since for sedimentary quartz some fluctuations were observed in the value of equivalent dose (De) for measurement temperatures at about 125 C (Kijek et al., 2013). In the standard dating procedure, the OSL signal is read out at 125 C. The presence of traps responsible for the 210 C peak has an influence on the OSL process, just as the traps responsible for the 110 C peak do. Hence the temperature of OSL measurement used in the SAR protocol should be selected with special care. In order to examine its influence on dating results, a series of measurements of De were carried out for various temperatures TOSL
http://dx.doi.org/10.1016/j.radmeas.2015.02.003 1350-4487/© 2015 Published by Elsevier Ltd.
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
2
during readout of the OSL signal in the SAR protocol, namely for: 25, 55, 90, 125, 160 and 195 C. For most samples significant fluctuations of De values were obtained as described below. TL measurements may provide a useful insight into carrier transfer processes during the steps of a SAR protocol and may explain the origin of De variations. TL was measured before each of seven initial steps of SAR. Changes of the TL curve shape in subsequent steps were observed and compared for all examined samples. Two groups of samples were distinguished which behave in a different way in the test of De dependence on temperature. 2. Experimental Quartz grains have been separated from 11 ceramics samples labeled: JK0, JK1, JK2, JK3, JK4, JK5, JAK2, JAK3, JAK4, JAK5, JAK6 in the way common for optically stimulated luminescence (OSL) dating. Mineral grains about 100e200 mm in diameter have been obtained for single aliquot measurements. Aliquots of 4 mm diameter have been prepared using Silkospray silicone oil to mount grains. Multiple grain (MG) SAR measurements as well as TL measurements have been carried out using a Risø TL/OSL reader. A blue LED stimulation module (470 nm, 80 mW cm2) has been used as an optical stimulation source for the MG method. For artificial irradiation, a 90Sr/90Y b source giving a dose rate of about 10 mGy s1 has been utilized. Another light source with a wavelength of 520 nm, similar to the wavelength of the laser used in the Risø Single Grain Attachment, has been also used for OSL measurements; it was fitted to a Risø TL/ OSL System TL-DA-12 equipped with an EMI 9235QA PMT and 2 mm Schott UG11-IRB filter. A strong Xenon lamp with a monochromator (180e1000 nm) and lightquide with a special adapter has been applied instead of a halogen lamp normally used for stimulation in TL-DA-12 readers. The photon flux used for stimulation has been 8.8 1016 cm2 s1 and the bandwidth of the stimulation was 26 nm. During irradiation a 90Sr/90Y source giving dose rate of 40 mGy/s was used. 3. Results 3.1. De dependency on the temperature of OSL measurement In order to determine De the single aliquot regeneration dose protocol (SAR) has been applied, which is typically used for age calculation. A preheat test has been performed before the actual study for each sample (from 180 C to 240 C, every 20 C) in order to find out the proper temperature of preheat for the standard SAR protocol. The whole SAR protocol has been repeated for six different temperatures TOSL of OSL readout: 25 C, 55 C, 90 C, 125 C, 160 C, 195 C. Sixteen aliquots of each sample have been used for estimating an individual De value for each temperature. The above-mentioned series of SAR protocol measurements have been applied for each of the 11 samples. Besides the observation of the dependence of De on the temperature, the repeatability of the results (the recycling test) and the quality of bleaching the OSL signal after a regeneration dose (the recuperation test) have been monitored. For each sample a dose recovery test was performed. In this case a known laboratory dose of the order of De has been reconstructed by applying the SAR protocol. Fig. 1 presents the results of De estimation for different temperatures of OSL measurement in SAR protocol. The shown values have been normalized to De obtained at 25 C. Fig. 2 shows similar results for the dose recovery test and Fig. 3 provides recycling test outcomes for each sample. The presented De values are obtained after discarding aliquots, for which De values differ significantly from the average
Fig. 1. The results of De estimation for different temperatures of the OSL measurement. Each value is normalized to De obtained at 25 C. For clarity of the figure the subsequent plots have been shifted upwards by the value 0.2. The grid lines are marked at values (0.9 þ i 0.2) and (1.0 þ i 0.2), where i ¼ 0, 1, …, 10 for each plot.
calculated for all 16 aliquots or for which the recycling ratio was outside the range 0.9e1.1. One can clearly see that higher temperatures of OSL measurement lead to a strong disturbance of De values for most samples. Although the character of these changes is sample-dependant, one can observe that the most intensive changes appear in the temperature region from 90 to 200 C, which is the characteristic temperature of intense TL peaks in quartz. There is, however, a group of samples, JK2, JK3, JAK2, JAK4 and JAK6, for which De variations are small and do not exceed 10%. Among them, only for four samples, JK3, JAK2, JAK4 and JAK6, have positive results been obtained also in the recovery and recycling tests. For the rest of the samples the De variations with measurement
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
Fig. 2. The results of recovery dose estimation for different temperatures of the OSL measurement. Each value is normalized to De obtained at 25 C. For clarity of the figure the subsequent plots have been shifted upwards by the value 0.2. The grid lines are marked at values (0.9 þ i 0.2) and (1.0 þ i 0.2), where i ¼ 0, 1, …, 10 for each plot.
temperature are larger than 10% and mostly bigger for higher temperatures. Some results of the recovery and recycling test for these samples, are nevertheless acceptable, as for example in the case of sample JK5 (both tests acceptable) or JK1, for which the recovery test has given positive results. Very good results for the recycling test have been obtained also for sample JAK5. 3.2. TL measurements One way to look inside OSL kinetics during SAR measurements is to observe TL curve changes after subsequent steps of this protocol.
3
For each investigated sample, TL curves were measured (with a heating rate 5 K s1) at each stage of the protocol, so a set of 7 TL curves was created: natural TL, natural TL after preheat (the temperature established by preheat test for OSL), natural TL after OSL measurement (40 s at 125 C), laboratory TL (after test dose), laboratory TL after preheat, laboratory TL after OSL measurement, laboratory TL after OSL bleaching at 280 C. Each TL curve has been measured for 3 fresh aliquots that individually have undergone adequate number of SAR steps. Fig. 4 illustrates the key findings from the analysis of the outcome of TL measurements. The natural as well as laboratory TL curves differ in the intensities of individual TL peaks (Fig. 4a, b). A peak about 200 C dominates in the natural curves. The peak at 110 C is also present before preheat. In the laboratory excited curves the peak at 110 C dominates and apart from the peak at 200 C there appears a peak at 160 C of a comparative intensity. Here, it should be stressed that mutual relations between the intensities of individual peaks which are seen in TL curves do not reflect the real relation between the concentrations of traps responsible for those peaks. The very low intensity of high-temperature peaks may be caused by thermal quenching (Wintle, 1975). As it was demonstrated recently, taking into account this phenomenon and correcting the TL curve reveals that the concentration of high-temperature traps can be comparable to the concentration of the traps responsible for the 110 C TL peak (Subedi et al., 2012). The analysis of TL curves distinguishes two groups of the investigated samples that correlate with the groups identified on the basis of De variability with the temperature of OSL measurement. In Fig. 4a natural curves for the representatives of both sample groups (JK3 and JAK5) are compared. In the case of samples JK2, JK3, JAK2, JAK4, and JAK6, the 200 C peak dominates, whereas the 110 C peak is hardly observed. In the remaining samples the intensity of the latter peak is similar to the intensity of the 200 C peak. As mentioned previously, in the laboratory excited TL curves (Fig. 4b), the 110 C peak prevails, but in the curves of the samples listed above high-temperature peaks can still be clearly seen, although they are about 6 times weaker. This is not the case for other samples. The intensity of the 110 C peak is about 50 times greater than the peaks at 160 and 200 C. Another TL feature that differentiates the two sets of samples is presented in Fig. 5. It is connected with the presence of the peak just above 200 C in the natural as well as laboratory TL curves measured after preheat and OSL stimulation. In the curves of samples JK0, JK2, JK3, JAK2, and JAK4 this peak is always noticeable, even after a preheat at 240 C (Fig. 5a), but in the other samples it is not present (Fig. 5b) after preheat at 200 C. The next difference is that the decrease in the intensity of the TL curve after the OSL stimulation is evident in the curves for the first sample set but for the remaining samples the decrease in the TL peak intensity is hard to notice or, as it is illustrated for sample JAK5, one can observe the peak increase after optical stimulation. 3.3. OSL component study Observing the fluctuations of De it is worth testing the changes of OSL curve composition with the OSL measurement temperature. In order to find the OSL signal components a procedure of fitting the sum of first order OSL curves to the experimental OSL curve has been applied. The non-negative least squares procedure available in MATLAB has been used. In all outcomes the fitting of first order components to the OSL curves was satisfactory (FOM (figure of merit) for fitting results are always less than 2%). An example of the outcomes of the fitting procedure (Fig. 6.) shows the component OSL curves and their sum, together with the OCS (optical crosssection) spectra, i.e. intensities obtained for individual OCS values
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
4
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
Fig. 3. Results of the recycling test for SAR protocol carried out with various temperatures of OSL measurement. For each temperature an average of the results for 16 aliquots is given.
Fig. 4. Natural (a) and b) laboratory excited (b) TL curves for samples JK3, JAK5 and JK0. The heating rate used in TL measurements was 5 K s1, the dose applied for TL excitation e 0.75 Gy. In part (a) a laboratory excited TL curve for sample JK3 is added in order to facilitate the shape comparison for the both kinds of curves.
from the range 1024e1014 cm2 presented in the form of a column plot. Fig. 7 presents the results of decomposition of the OSL curves measured at different temperatures (25, 55, 90, 125, 160 and 195 C) into first-order components. Sample JAK5 was chosen as a representative of the samples that manifest significant variations of De. The OSL curve analysis was carried out for naturally and laboratory (beta dose of about 1 Gy) irradiated samples. The experiment for an individual temperature was repeated for three aliquots. The results were similar for natural and laboratory curves; hence they are presented only for the first case. Five components were found for all the tested temperatures. The most conspicuous observation concerns the behavior of the slow component. Its intensity fluctuates significantly at temperatures above 90 C. The intensities of all other components change their mutual relation in this temperature range, but the changes are not as radical as for the slow component. Simultaneously, constant decrease of the intensities of all components except for the slow one starts already above 60 C. This effect can be explained by the thermal quenching of luminescence. A similar series of experiments has been made for stimulation with light of 520 nm wavelength (bandwidth 26 nm). The outcomes
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
5
Fig. 5. TL curves (rate 5 K s1) obtained after the subsequent steps of SAR protocol: natural signal after preheat (black dash line) and after the OSL measurement (black solid line); laboratory TL after the preheat (light gray dash line) and after OSL measurement (light gray solid line). Examples are given for two samples e JK3 (a) and JAK5 (b) that are the good representatives of two sample sets that can be distinguished because of their TL features.
for natural OSL are shown in Fig. 8. For this lower stimulation energy the OCS spectra are less complex than for the stimulation with light at 470 nm. Three components are active in the whole range of temperatures. The OCS values for 520 nm stimulation are consistently lower than those for 470 nm with dependence of the OCS on the stimulation energy. The three components observed for 520 nm stimulation are probably the two fast and medium 1 components observed during the stimulation by blue diodes. Their intensities decrease monotonically in a similar way as was observed for 470 nm stimulation. The slow component, so intensive and extraordinary for 470 nm, is not present or rather sinks in the background. 4. Discussion The temperature of the OSL measurement in the SAR protocol applied for sediment dating was determined on the basis of investigations carried out for quartz grains from sediment (Murray
Fig. 6. The results of fitting the sum of the first order curves (crosses) to the experimental OSL curve (solid black line) and OSL components for a chosen aliquot. The bottom part e the difference between the curve obtained from fitting (sum of the OSL component) and the experimental OSL curve. Inset: the OCS spectra estimated by fitting procedure for all 3 aliquots measured in the same conditions. Spectra for the individual aliquots are presented in different colors.
Fig. 7. Results of OSL component analysis by fitting the sum of firsteorder OSL curves to an experimental curve. The dependence of the outcomes of fitting: the OCS values (a) and the initial concentration of electrons in traps (b) on the temperature of OSL measurement for the stimulation light of the wavelength of 470 nm (blue LEDs used in MG SAR). Each point on the graphs represents the average of 3 aliquots.
and Wintle, 1998). Reading out the OSL signal at a temperature about 125 C should keep the 110 C TL trap empty, which is responsible for slowing down the OSL process. In quartz from bricks that were annealed for a long time at high temperature during the brick production process, there are also traps that give TL peaks at about 160 and 200 C. These traps, contrary to what might be thought on the basis of the shape of the TL curve, have concentrations comparable to the concentration of the 110 C TL trap. As mentioned earlier, the low intensities of the peaks next to the 110 C peak are a consequence of the strong thermal quenching effect in quartz. The TL curves presented here for samples JK3 and JAK5 (Fig. 4) allow one to suspect that the concentration of 160 C TL traps can be even higher than the concentration of the 110 C TL traps. Therefore, the idea to avoid the slowing down of OSL process by shallower traps by keeping them empty should be also addressed to the 160 C and 200 C traps. This would require much higher temperature of OSL measurement in the SAR protocol. As the present study shows, this is not recommended because of the high fluctuations of the De values obtained in such readout conditions observed for many samples. Another disadvantage of the high OSL readout temperature is the impact of thermal quenching. It is, however, hard to attribute the poor results of De estimation for higher temperatures to this effect alone. The TL experiments allowed one to divide the investigated samples into two groups. The samples from the first group (JK2, JK3, JAK2, JAK4 and JAK6) are characterized by a high intensity of the 200 C TL peak, comparable to the intensity of the 160 C TL peak or higher (Fig. 4b, sample JK3), the clear optical bleaching of the TL signal in the range 300e400 C during the SAR protocol (Fig. 5a) and the presence of the 200 C TL peak in the TL signal even after the preheat at 240 C. This group demonstrates weak De fluctuations (below 10% of the value obtained at 25 C) and mostly for the
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003
6
ska / Radiation Measurements xxx (2015) 1e6 N. Kijek, A. Chruscin
because of the 200 C TL traps) that may lead to significant fluctuations of De values with temperature and poor outcomes of the recycling and recovery tests. Therefore, one should consider changing the OSL measurement temperature in the SAR protocol to a significantly lower one, below the range of dominant TL peaks. In the range RT e 60 C the De fluctuations are less frequent for most samples. Simultaneously the recovery test for this temperature range gave good results for most of the samples. The investigation of the OSL components for 520 nm stimulation shows that this lower stimulation runs more stable (Fig. 8). Only three OSL components are active and the OCSs increase with temperature according to their normal behavior. Applying the 520 nm stimulation should be more intensively tested in application for OSL dating of ceramics. 5. Conclusions In quartz from ceramics the De value can strongly depend on the temperature of OSL measurement. The most significant fluctuations appear at temperatures above 90 C and are linked to poor results in the recycling test. The above effects are not so clearly evident in dose recovery experiments. In quartz from ceramics, beyond the 110 C TL traps there are also 160 and 200 C TL traps having very high concentrations, hence the 125 C temperature for OSL measurement in the SAR protocol does not assure the exclusion of the shallow traps from the OSL process. A reasonable solution to this problem may be applying a temperature below the range of the dominant TL peaks, in the SAR protocol for OSL measurements. Acknowledgments Fig. 8. Results of OSL component analysis by fitting the sum of firsteorder OSL curves to an experimental curve. The dependence of the outcomes of fitting: the OCS values (a) and the initial concentration of electrons in traps (b) on the temperature of OSL measurement for the stimulation light of the wavelength of 520 nm. Each point of the graphs represents the average from the results for 3 aliquots.
higher temperatures. The rest of samples have a very strong 160 C TL peak in comparison to the 200 C TL peak, and their TL signal in the range 300e400 C is not bleached during the optical stimulation in the SAR protocol (Fig. 5a). Presumably, traps deeper than the traps responsible for the TL signal in the range 300e400 C are the source of the OSL signal. It is, however, not clear why in the samples having lower concentration of the 160 C TL traps (JK2, JK3, JAK2, JAK4 and JAK6) the bleaching of TL in the range 300e400 C can be observed. The shallow TL traps have an impact on the individual components of the OSL signal (Fig. 7). Normally, the OCS values should increase with temperature. The activity of shallower traps disturbs this normal behavior in the range above 60 C. The OCSs fluctuate. It is most evident for the slow component. Intensities of all components decrease because of thermal quenching. The relation of the intensities of fast and slow components may be the most important for the quality of De estimation. The decrease of slow component intensity with temperature could be treated as a positive effect, however at the same temperature the ratio of slow component's OCS value to fast component's OCS value increases, so the part of the OSL related to the slow component in the total OSL signal stays as it was at lower temperatures. Summing up, the measurement of OSL at 125 C in the course of the SAR protocol applied to quartz from bricks seems to have no advantages such as increasing the total luminescence intensity or improving the ratio of the part of the OSL originated from the fast component to the slow component's part. Additionally the OSL kinetics in this temperature region is very complex because of the high concentration of 110 C and 160 C TL traps (and may be also
This work has been done thanks to the help of the Centre for Modern Interdisciplinary Technologies, Nicolaus Copernicus University in Torun, ul. Wilenska 4, 87-100 Torun, Poland (e-mail: [email protected]) and the grant No. 1765-F obtained from Nicolaus Copernicus University. References Bailiff, I.K., Petrov, S.A., 1999. The use of the 210C TL peak in quartz for retrospective dosimetry. Radiat. Prot. Dos. 84, 551e554. Bailiff, I.K., 2007. Methodological developments in the luminescence dating of brick from English late-medieval and post-medieval buildings. Archaeometry 49, 827e851. , M., 2010. An intercomparison Blain, S., Bailiff, I.K., Guibert, P., Bouvier, A., Bayle study of luminescence dating protocols and techniques applied to medieval brick samples from Normandy (France). Quat. Geochronol. 5, 311e316. ska, A., Oczkowski, H.L., Przegie˛ tka, K., 1996. Trap spectra of annealed Chruscin quartz. Acta Phys. Pol. A 89, 555e568. Goedicke, C., Slusallek, K., Kubelik, M., 1981. Thermoluminescence dating in architectural history: Venetian villas. J. Soc. Archit. Hist. 40 (3), 203e217. Guibert, P., Bailiff, I.K., Blain, S., Gueli, A.M., Martini, M., Sibilia, E., Stella, G., Troja, S.O., 2009. Luminescence dating of architectural ceramics from an early medieval abbey: the St Philbert Intercomparison (Loire Atlantique, France). Radiat. Meas. 44 (5e6), 488e493. Kennedy, G.C., Knopf, L., 1960. Dating by thermoluminescence. Archaeology 13, 147e148. ska, A., Przegietka, K.R., 2013. On the dependence of equivalent Kijek, N., Chruscin dose on temperature of OSL measurements for sediment quartz grains and its implication in dating practice. Radiat. Meas. 56, 252e256. Murray, A.S., Wintle, A.G., 1998. Factors controlling the shape of the OSL decay curve in quartz. Radiat. Meas. 29, 65e79. Petrov, S.A., Bailiff, I.K., 1997. Determination of trap depths associated with TL peaks in synthetic quartz (350e550 K). Radiat. Meas. 27, 185e191. Subedi, B., Polymeris, G.S., Tsirliganis, N.C., Pagonis, V., Kitis, G., 2012. Reconstruction of thermally quenched glow curves in quartz. Radiat. Meas. 47, 250e257 (of the Royal A). Wintle, A.G., 2008. Fifty years of luminescence dating. Archaeometry 50, 276e312. Wintle, A.G., 1975. Thermal quenching of thermoluminescence in quartz. Geophys. J. R. Astronomical Soc. 41, 107e113. Wintle, A.G., Murray, A.S., 2006. A review of quartz optically stimulated luminescence characteristics and their relevance in single aliquot regeneration dating protocols. Radiat. Meas. 41, 369e391.
ska, A., Equivalent dose of quartz originating from ceramics obtained by OSL SAR method e Please cite this article in press as: Kijek, N., Chruscin Tests of protocol parameters, Radiation Measurements (2015), http://dx.doi.org/10.1016/j.radmeas.2015.02.003