Optically stimulated luminescence of Al2O3

Optically stimulated luminescence of Al2O3

PII: Radiation Measurements Vol. 29, No. 3±4, pp. 391±399, 1998 # 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain S1350-4487(...

369KB Sizes 4 Downloads 65 Views


Radiation Measurements Vol. 29, No. 3±4, pp. 391±399, 1998 # 1998 Elsevier Science Ltd. All rights reserved Printed in Great Britain S1350-4487(98)00061-4 1350-4487/98 $19.00 + 0.00

OPTICALLY STIMULATED LUMINESCENCE OF Al2O3 M. S. AKSELROD,1,2 A. C. LUCAS,2 J. C. POLF1 and S. W. S. McKEEVER1* Department of Physics, Oklahoma State University, Stillwater, OK 74078-3072, USA and 2Stillwater Sciences LLC, 206 Abbey Ln., Stillwater, OK 74075-1933, USA


AbstractÐAnion-de®cient aluminum oxide doped with carbon (Al2O3:C) is not only an extremely sensitive thermoluminescence (TL) material, but is well suited to optically stimulated luminescence OSL applications due to a high cross-section for interaction of light with radiation-induced trapped charge. Several di€erent OSL readout protocols have been suggested, including pulsed OSL (POSL), and ``delayed'' OSL (DOSL). This paper examines the properties of Al2O3:C for application using these two readout protocols. The POSL technique utilizes the prompt luminescence that results from the direct recombination of released charge carriers at luminescence sites (F-centers in Al2O3:C). Following a pulse of stimulation light using a laser, the POSL signal is observed to decay with a temperature-independent lifetime of 035±36 ms. The DOSL signal, on the other hand, utilizes the temperature-dependent signal resulting from the capture of released charge carriers by shallow traps. The decay of the luminescence component after the stimulating pulse has a lifetime of several hundred ms, depending upon temperature. The dependence of the DOSL signal on readout temperature can be explained in terms of the involvement of the shallow traps in the process. However, the intensity (not the lifetime) of the POSL signal is also slightly temperature dependent. It is conjectured that this may be caused by a thermally assisted optical detrapping process involving localized excited states. Di€erent forms of Al2O3:C are examined. By modifying both the concentration and energy distribution of the shallow traps material optimized for DOSL applications can be engineered. In contrast, the best material for POSL is grown with no shallow traps. The integrated light output in a typical POSL measurement is approximately a factor of 7±8 greater than that of DOSL, even for a DOSL-quality sample. # 1998 Elsevier Science Ltd. All rights reserved


temperature. The relatively narrow band gap possessed by these materials dictates that the traps used to store the latent dose information are too shallow to store charge for signi®cant periods at room temperature, as is evidenced by the fact that the charge can be eciently detrapped optically using low energy IR stimulation. These phosphors also have a very high e€ective atomic number and, as a result, they exhibit a strong energy dependence which is unacceptable for personal dosimetry. Several research groups have tried to use optical stimulation as a dosimetric tool by optically transferring charge carriers from deep traps to shallow traps and then monitoring the phosphorescence at room temperature. Several phosphors were used in these studies, including BeO (Tochilin et al., 1969; Rhyner and Miller, 1970), CaF2:Mn (Bernhardt and Herforth, 1974; Hanniger et al., 1982) and CaSO4:Dy (Pradhan, 1977; Pradhan and Bhatt, 1981). Each of these materials, however, exhibited a relatively low sensitivity. Furthermore, since the method relies upon the behavior of traps which are unstable at ambient temperatures it is intrinsically sensitive to the actual ambient temperature during the readout process. This OSL readout mode is referred to as ``Delayed'' OSL (DOSL) and com-

Optically stimulated luminescence (OSL) was ®rst suggested as a dosimetry tool in the 1950s and 1960s (Antonov-Romanovskii et al., 1956; BraÈunlich et al., 1967; Sanborn and Beard, 1967). Although the technique is now actively used by the archaeological and geological dating community as a method of equivalent dose determination in natural materials (e.g. Wintle, 1997), following the introduciton of the method in this area by Huntley et al. (1985), the continued use of OSL in personal radiation dosimetry has been less extensively reported since the early suggested uses noted above. The main reason for this has perhaps been the lack of a suitable luminescent material having a high sensitivity to radiation, a high optical stimulation eciency, a low e€ective atomic number, and good fading characteristics (i.e. a stable luminescence signal at room temperature). Early dosimetric applications used MgS, CaS, SrS and SrSe doped with di€erent rare earth elements such as Ce, Sm and Eu (BraÈunlich et al., 1967; Sanborn and Beard, 1967; Rao et al., 1984). Although these materials possess a high sensitivity to radiation and a high eciency under IR stimulation, at a wavelength around 1 mm, they su€er from signi®cant fading at room *To whom all correspondence should be addressed. 391

M. S. AKSELROD et al.


mercial dosimetry systems have now been developed based on this method (Yoder and Salasky, 1997). A modi®cation of the OSL technique, called Pulsed Optically Stimulated Luminescence (POSL), was recently reported using crystalline Al2O3:C (Markey et al., 1995; McKeever et al., 1996). The essence of the technique is to expose an irradiated sample to a pulsed light source and to detect the emitted luminescence between pulses, but not during the pulses. This arrangement allows one to use less ®ltration to discriminate between the stimulation light and the luminescence, at the same time allowing one to bias against the slow phosphorescence processes which make up the main signal in DOSL measurements. These features grant the POSL technique both a high sensitivity and a weaker temperature dependence compared with the DOSL method. The timing parameters for the pulsing and the luminescence measurement are carefully selected to match the material being examined. A critical property in choosing both a suitable material and the appropriate timing parameters is the luminescence lifetime for the material under study. In the studies by Markey et al. (1995) and McKeever et al. (1996) the phosphor chosen was Al2O3:C, initially developed for TL applications (Akselrod et al., 1993). Extensive studies have shown that this material possesses a high optical sensitivity in that charge trapped at thermally stable defects has a large photoionization cross-section for detrapping (Akselrod et al., 1990; Walker et al., 1996; Bùtter-Jensen and McKeever, 1996). The goal of this paper is to describe additional recent studies on the suitability of Al2O3:C for use in OSL dosimetry, particularly POSL and DOSL. Emphasis is on the availability of di€erent forms of this material selected to optimize its use in these two OSL procedures, and to discuss the relative sensitivity of the two methods. 2. PROCEDURES 2.1. POSL versus DOSL The POSL method consists of stimulating irradiated samples with a train of pulses from a laser. The luminescence from the samples is detected after the pulses, but not during them. By ignoring the luminescence output during the stimulation pulses we prevent problems of having to discriminate between the intense stimulating laser light and the weak emitted luminescence. In this way we are able to record the luminescence output without the use of heavy ®ltration to remove the laser light. Appropriate ®ltration to prevent damage to the photomultiplier tube is all that is required. POSL measurements require upon a knowledge of the luminescence lifetime of the material under study. For the Al2O3:C samples used in this study, the luminescence arises from the relaxation of

excited F-centers with a broad emission peaking at 0420 nm and a lifetime t of 035±36 ms. OSL signals, therefore, consist of a ``rapid'' component due to those charges which are optically detrapped and undergo prompt recombination without ®rst being re-trapped in shallow traps. In addition to this temperature-independent component there is a temperature-dependent component with a long lifetime of more than a few hundred ms at room temperature. This component is due to the re-trapping of the charge by shallow traps (Markey et al., 1995). Thus, for these short data acquisition times (i.e. times signi®cantly shorter than t) we monitor the strong, prompt, temperature-independent OSL component which, at these short times, is stronger than the delayed, temperature-dependent component. This is the essence of the POSL measurement. Alternatively, by delaying the data acquisition after the period of laser stimulation until such time as the rapid component has decayed (i.e. for times >ca. 3t) and then acquiring the data over a long enough period such that one can record all, or most, of the slow component, we bias the data collection in favor of that charge which underwent recombination only after being re-trapped in the shallow traps. Thus, only the slow component is monitored and this is the DOSL signal. Thus, the two processes (POSL and DOSL) can be easily separated using time discrimination. The distinction between the POSL process and the DOSL process can be further understood with reference to Fig. 1 which shows a schematic representation of the two processes on a ¯at-band energy band diagram. The transitions which give rise to the prompt POSL emission and to the delayed DOSL emission are both illustrated.

2.2. Equipment The POSL measurements were recorded using principles already described in previous publications (McKeever et al., 1996). The output power was continuously monitored and used to correct the data for laser power ¯uctuations. All timing parameters were controlled by a computer. Sample irradiations were performed at room temperature using a 90Sr/90Y beta source. Thermoluminescence (TL) measurements were performed in a vacuum in a liquid nitrogen cryostat. The samples were irradiated inside the cryostat using a 137Cs source at 200 K, and were heated after irradiation to a temperature of 500 K at a heating rate of 0.2 K/s. The photomultiplier output (integrated current) was collected by a computer for analysis. The samples were grown by Stillwater Sciences, LLC in a reducing atmosphere in the presence of carbon. Samples of three major types were selected for the study. These were: (i) TLD-quality ma-



3. RESULTS 3.1. Thermoluminescence

Fig. 1. A schematic energy band model showing the direct transitions which lead to the prompt luminescence signal utilized in POSL, and the transitions associated with trapping of the released charge at the shallow traps. The latter cause a delayed luminescence signal, which is monitored during DOSL measurements.

terials, similar to TLD-500 K available previously (Akselrod et al., 1993). (ii) POSL-quality samples, grown without the presence of shallow traps in order to enhance the prompt luminescence. (iii) DOSL-quality samples, grown in such a way as to optimize the concentration and energy distribution of the shallow traps so as to produce a large, delayed luminescence component. All samples were single crystals, of dimensions 5 mm diameter and approximately 1 mm thick.

Thermoluminescence glow curves for each of the three types of sample are shown in Fig. 2. In each case the dose delivered was approximately 150 mGy. Each curve is characterized by glow peaks in three main temperature regions: (A) 0230±280 K; (B) 0280±320 K; and (C) 0400±480 K. The glow peaks are described here in terms of temperature regions, rather than peak positions, because the positions varied slightly from sample type to sample type and, as will be shown, they are caused by a distribution of trapping statesÐi.e. the peaks are not caused by charge release from traps with discrete trap depths. Previous analyses (e.g. Markey et al., 1995; Walker et al., 1996) have demonstrated the ®rst-order nature of the TL peaks from Al2O3:C at low doses, in contrast to the claims of earlier analyses (Kitis et al., 1994; Kortov et al., 1994). An additional complication when analyzing the glow peaks in region (C) is the presence of thermal quenching, which must be accounted for. As with OSL, the emission in TL is also from the relaxation of excited F-centers and the eciency of this emission thermally quenches for temperatures in the region of peak (C). However, the main concern of the present study is with the glow curve in regions (A) and (B), since these govern the time response of the DOSL signal of interest in this paper. Hence, thermal quenching e€ects were not an issue.

Fig. 2. Glow curves from (1) POSL-quality, (2) TLD-quality, and (3) DOSL-quality Al2O3:C samples irradiated at 200 K and heated at 0.5 K/s. TL is observed in three temperature regions: (A) 0230± 280 K; (B) 0280 K±320 K; and (C) 0400 K±480 K. The latter is due to the trapping of charge at stable traps and it is this trap population that is utilized in both TL and OSL dosimetry.

M. S. AKSELROD et al.


We note that the POSL-quality material does not have either of the shallow trap distributions (regions (A) and (B)), whereas both the TLD-quality and the DOSL-quality material display signi®cant TL in these temperature regions. To gain an estimate of the trap depths distributions in regions (A) and (B) we employed a simpli®ed deconvolution routine, as described by Agersnap Larsen (1997) and by Colyott (1997). In brief, one starts from a two-dimensional Fredholm equation, namely: ZZ I…T† ˆ g…E; s†IRW …E; s; T†dEds …1† where g(E,s) is the distribution in E and s space of the weighting function for the Randall±Wilkins (RW) 1st-order equation: ZT IRW …E; s; T† ˆ s expfÿE=kTg …ÿs=b† expfÿE=kygdy: To

…2† However, the RW function shape and position are less sensitive to s than they are to E. Furthermore, if only one heating rate is used, [1] requires the determination of a two-dimensional function g(E,s) from a one-dimensional data set I(T). Hence, since we only require in the present analysis a semi-quantitative estimate for the energy distributions in order to compare the di€erent samples, we have simpli®ed [1] to the one-dimensional case, namely: Z …3† I…T† ˆ g…E†IRW …E; s; T†dE by assuming that s is a constant value across the glow curve region of interest. In the above equations E is the trap depth, s is the frequency factor, T and y are temperature, k is Boltzmann's constant and b is heating rate. In our analysis [3] is further approximated by a summation, thus: I…T† ˆ

E2 X

g…E†IRW …E; s; T†



where the summation limits (E1 and E2) and the interval size DE (i.e. the number of summed components, and the energy resolution), and the value of s, are all preselected by the operator. Thus, the glow curve is described as the sum of several functions, each of the form of the Randall±Wilkins equation and each with the same value of s, and ®xed but di€erent values of E and g(E). The summation is then ®t to the experimental glow curve using a Marquardt±Levinson least-squares ®tting routine, with the g-values as variables. The result is an energy distribution (a ``density of states'') corresponding to a ®xed value for s. In the analyses presented here, s was ®xed at 1014 sÿ1. The result of the deconvolution process for the TLD-quality and DOSL-quality materials is shown

in Fig. 3. Here the deconvolution result for the glow curve in region (A) is shown in Fig. 3(a), and that of region (B) is shown in Fig. 3(b). Both results indicate that the traps giving rise to the TL in these regions are distributed in energy. Furthermore, by comparing the results from the DOSL-quality material with those of the TLD material we see that both the energy of the distribution maximum and the width of the energy distribution are di€erent in each case. In particular, we see that the distribution is shifted to higher energies for the DOSL material. Even if it is argued that the energy values obtained are inaccurate since we assumed a ®xed value for s, the point of the analysis is to reveal that the glow curves are described by an energy distribution, and not discrete states. Furthermore, the point is clearly made that the distributions, and not just the size of the peaks, is di€erent in the DOSL quality material compared with the other material types. It should be noted that all attempts to ®t these glow curves using either the RW equation, or the conventional equations for non-®rst-order kinetics (McKeever, 1985), but assuming traps with single-valued trap depths, failed. High quality ®ts could only be obtained by assuming a distribution of trapping energies.

3.2. POSL and DOSL Figure 4 shows the POSL and DOSL signals for the three samples. In each case the samples had been irradiated with a dose of 10 mGy of 90Sr/90Y beta particles at room temperature and then exposed to the 532 nm pulsed laser beam. During the ®rst 1000 ms in Fig. 4 the laser beam was incident on the sample. In this period each datum corresponds to the measured POSL from the samples between each laser pulse, as described in the experimental section. We observe an initial increase in POSL intensity as an equilibrium population of excited luminescence sites (i.e. F-centers) is established. After this we see a slow, gradual decrease in intensity with continued laser exposure as the source traps are depleted. Extensive earlier measurements showed that the traps being depleted during laser exposure are the same traps as those that cause the glow peak in region (C), Fig. 2 (Markey et al., 1995; Walker et al., 1996). At the end of 1 s the laser stimulation was stopped and the luminescence decays. This component represents the DOSL signal, which follows the last laser stimulation pulse. The time dependent behavior of this portion of the signal is signi®cantly di€erent for the three samples. In the POSL-quality material the luminescence signal decays exponentially with a time constant of 035±36 ms, corresponding to the lifetime t of the excited F-centers. This component is temperature independent. The TLD-quality material displays an initial rapid



Fig. 3. Trap energy distributions calculated from a deconvolution of the glow curves in regions (A) (®g. (a)) and (B) (®g. (b)), using the method described in the text. The distribution for the DOSL-quality material is clearly shifted to higher energies.

decay, also with a temperature-independent lifetime of 35±36 ms, and then shows a slower decay with an approximate lifetime of 0350 ms at this temperature. This component is temperature dependent in the temperature range (below 0370 K, beyond which thermal quenching of the luminescence occurs). The third sample, i.e. the DOSL-quality material, shows a small, initially rapid, decay followed by a slow decay with a temperature dependent lifetime of 0660 ms, although the decay is not

quite exponential (the decay curve in the ®gure is non-linear on the semi-log plot). Overall, we have observed decay times varying from 500 ms to as long as 1000 ms for this slow component, depending upon the particular sample used. Some decays are clearly non-exponential. To understand this behavior we refer to the TL glow curves of Fig. 2 and the energy spectra of Fig. 3. Here we saw that the POSL-quality material does not possess any shallow traps and therefore

M. S. AKSELROD et al.


Fig. 4. POSL and DOSL signals from the three types of Al2O3:C sample: (1) POSL-quality; (2) TLDquality; (3) DOSL-quality. During the period 0±1000 ms the pulsed laser stimulation is incident on the sample. The POSL signal is measured between the pulses. After the last pulse (at 1000 ms) the OSL signal decays. The POSL-quality sample displays only the rapid 35±36 ms decay associated with the lifetime of the excited F-centers. The other two samples each show, in addition, a longer lived component, with lifetimes of several hundred ms, due to the trapping of charge in shallow traps.

the slow, temperature-dependent DOSL decay is entirely missing. In contrast, the other two sample types each have a considerable shallow trap concentration (as evidenced by the intense TL in regions (A) and (B)), and signi®cant DOSL signals. We note, however, that it is not just the number of shallow traps that is important, but also the energy distribution of these traps. Not only is the DOSL signal from the DOSL-quality sample stronger than it is from the other two samples, but the DOSL decay is longer, indicating a contribution from deeper trapping states. This is consistent with the semiquantitative trap depth distributions displayed in Fig. 3 where we observed that the DOSL material is characterized by a distribution of traps with deeper trap depths. 3.3. Temperature dependence In Fig. 5 we illustrate the temperature dependence of the POSL and DOSL signals. The temperature of interest here is the temperature of readout during POSL or DOSL measurement. Figure 5(a) shows the readout temperature dependence of the DOSL and POSL signals for the POSL-quality sample. The POSL signal is de®ned as the integrated OSL between 0 and 1000 ms. The DOSL signal is de®ned as the integrated OSL between 1200 and 5000 ms. Note that we do not take account of the data in the interval from 1000 (at the end of the last laser pulse) to 1200 ms. This delay allows the prompt component to completely

decay so that we are only monitoring that OSL component which is delayed by the action of the shallow traps. The ®gure shows a slight increase in the POSL output as a function of temperature over the temperature range 245±375 K. Clearly, there is no DOSL signal from this material, since there are no shallow traps. Markey et al. (1996) and Agersnap Larsen (1997) have examined the temperature dependence of cwOSL from TLD-500 and have modeled possible causes, including the role of shallow traps and the possibility of a thermally assisted transition to the delocalized band. In the latter process trapped electrons are assumed to be excited to a localized state just below the conduction band, similar to the mechanisms proposed to explain thermally assisted IR-stimulated OSL from feldspar. Thermal excitation then causes a transition from the excited state into the band, from where recombination can occur leading to OSL. The modeling studies of Markey et al. (1996) and Agersnap Larsen (1997), which are based on the ¯at band energy band diagram and rate equation analysis, suggest that either transitions into and out of shallow traps, or thermally assisted optical detrapping into the delocalized bands, could give rise to temperature dependencies of the type observed in those studies. However, in the POSL-quality samples discussed here there are no shallow traps and this leads to the possibility that the POSL temperature dependence observed is in fact caused by a thermally assisted process involving an excited localized state.



Fig. 5. (a) POSL from POSL-quality material as a function of the readout temperature. A weak temperature dependence is observed, possibly due to a thermal assisted optical detrapping process. (b) POSL and DOSL from DOSL-quality material. Each displays a strong temperature dependence due to the involvement of shallow traps.

In Fig. 5(b) we see the temperature dependence of the POSL and DOSL signals from the DOSLquality material. Clearly, the temperature dependencies are more substantial and complex than from the POSL-quality sample. For POSL, a step-like increase is observed at temperatures which approximately correspond to the temperatures of the TL signals (A) and (B) seen in Fig. 2. As each of these trapping states become unstable, the collected OSL emission increases correspondingly. For the DOSL

signal, we see an initial increase in output, followed by a decrease as the temperature increases further. At low temperatures, the charge in the shallow traps is stable and remains trapped over the course of the measurement period (i.e. from t = 1200 ms to t = 5000 ms). As a result, this trapped charge does not contribute to luminescence and there is little or no DOSL signal at these temperatures. As the temperature increases, however, the lifetime in the shallow traps decreases and the DOSL signal gets

M. S. AKSELROD et al.


progressively larger. Upon further increase in temperature, the shallow traps no longer become e€ective charge localizers. The lifetime falls to values less than the prompt luminescence lifetime of the luminescence centers (i.e. 35±36 ms) and once again no DOSL signal is observed. 3.4. Sensitivity Expressing the POSL signal once again as the integrated light output between 1 and 1000 ms, and the DOSL output as the integrated signal between 1200 and 5000 ms, we ®nd that the POSL signal is some 7±8 times more intense that the DOSL signal. That is, some 84±87% of the recombination luminescence results from direct transition of the released charge directly to the recombination centers, rather than via the shallow traps. We conclude that most of the charge optically released during laser excitation undergoes prompt recombination. This indicates that POSL is a substantially more sensitive method for dosimetry than DOSL even in the best of the DOSL-optimized material.

4. SUMMARY In this study we have examined the fundamental time response of the POSL and DOSL signals from Al2O3:C. During crystal growth the material can be optimized to the particular application. For POSL measurements, shallow traps can be removed from the material, leading to prompt luminescence only. This signal is caused by direct recombination of the optically released charges at the luminescence sites (F-centers) and is characterized by a lifetime of 035±36 ms. For DOSL measurements, a strong and long-lived DOSL signal can be produced by growing the material with large numbers of shallow traps and with a wide trap depth distribution. This results in a temperature-dependent component with a lifetime of several hundred milliseconds. The POSL signal, from POSL optimized material, is weakly temperature dependent and is possibly due to a thermally assisted optical excitation process. Both the POSL signal and the DOSL signal from DOSL-quality material show a strong temperature dependence, due to the role played by the shallow traps. The POSL signal is signi®cantly more sensitive than DOSL, and is potentially a much more useful dosimetric tool. AcknowledgementsÐWe are grateful for helpful discussions with Dr L.E. Colyott, especially regarding assistance with the deconvolution of the TL glow curves. This research was sponsored by the Oklahoma Center for the Advancement of Science and Technology, through the Applied Research program, under grant AR6-005-5156.

REFERENCES Agersnap, Larsen N. (1997) Dosimetry based on thermally and optically stimulated luminescence. Ph.D. thesis. Niels Bohr Institute, University of Copenhagan. Akselrod, M. S., Kortov, V. S., Kravetsky, D. J. and Gotlib, V. I. (1990) Highly sensitive thermoluminescence anion-defective a-Al2O3:C single crystal detectors. Radiat. Prot. Dosim. 32, 15±20. Akselrod, M. S., Kortov, V. S. and Gorelova, E. A. (1993) Preparation and properties of Al2O3:C. Radiat. Prot. Dosim. 47, 159±164. Antonov-Romanovskii, V. V., Keirum-Marcus, I. F., Poroshina, M. S. and Trapeznikova, Z. A. (1956) IR stimulable phosphors. In: Conference of the Academy of Sciences of the USSR on the Peaceful Uses of Atomic Energy. Moscow, 1955. USAEC Report AEC-tr-2435 (Pt. 1) pp. 239±250. Bernhardt, R. and Herforth, L. (1974) Radiation dosimetry by optically stimulated phosphorescence of CaF2:Mn. In: Proc. of 4th Int. Conf. On Luminescence Dosimetry. Krakow, Poland, pp. 1091± xxx. Bùtter-Jensen, L. and McKeever, S. W. S. (1996) Optically stimulated luminescence dosimetry using natural and synthetic materials. Radiat. Prot. Dosim. 65, 273±280. BraÈunlich, P., Shafer, D. and Sharmann, A. (1967) A simple model for thermoluminescence and thermally stimulated conductivity of inorganic photoconducting phosphors and experiments pertaining to infraredstimulated luminescence. Proc. 1st Int. Conf. on Luminescence Dosimetry. Stanford, June 1965, USAEC. 57±73. Colyott, L. E. (1997) Dosimetric properties of a-Al2O3:C exposed to ionizing and non-ionizing radiation. Ph.D. thesis. Oklahoma State University. Hanniger, J., Horlbeck, B., Hubner, K. and Prokert, K. (1982) The evaluation of CaF2:Mn-polyethylene detectors with the aid of the optically stimulated luminescence (OSL). Nucl. Instr. Meth. 204, 209± 212. Huntley, D. J., Godfrey-Smith, D. L. and Thewalt, M. L. W. (1985) Optical dating of sediments. Nature 313, 105±107. Kitis, G., Papadopoulus, J. G., Charalambous, S. and Tuyn, J. W. (1994) The in¯uence of heating rate on the response and trapping parameters of a-Al2O3:C. Radiat. Prot. Dosim. 55, 183±190. Kortov, V. S., Milman, I. I., Kirpa, V. I. and Lesz, J. (1994) Some features of a-Al2O3:C dosimetric thermoluminescence crystals. Radiat. Prot. Dosim. 55, 279±283. Markey, B. G., Colyott, L. E. and McKeever, S. W. S. (1995) Time-resolved optically stimulated luminescence from a-Al2O3:C. Radiat. Meas. 24, 457±463. Markey, B. G., McKeever, S. W. S., Akselrod, M. S., Bùtter-Jensen, L., Agersnap, Larsen N. and Colyott, L. E. (1996) The temperature dependence of optically stimulated luminescence from a-Al2O3:C. Radiat. Prot. Dosim. 65, 185±190. McKeever, S. W. S. (1985) Thermoluminescence of Solids. Cambridge University Press, Cambridge. McKeever, S. W. S., Akselrod, M. S. and Markey, B. G. (1996) Pulsed optically stimulated luminescence dosimetry using a-Al2O3:C. Radiat. Prot. Dosim. 65, 267±272. Pradhan, A. S. (1977) Radiation dosimetry by photostimulated luminescence of CaSO4:Dy. Int. J. Appl. Radiat. Isotopes 28, 534±535. Pradhan, A. S. and Bhatt, R. C. (1981) Photostimulated luminescence and thermoluminescence in CaSO4:Dy. Phys. Stat. Sol. (a) 68, 405±411.

OPTICALLY STIMULATED LUMINESCENCE OF Al2O3 Rao, R. P., de Murcia, M. and Gasiot, J. (1984) Optically stimulated luminescence dosimetry. Radiat. Prot. Dosim. 6, 64±66. Rhyner, C. R. and Miller, W. G. (1970) Radiation dosimetry by optically stimulated luminescence of BeO. Health Phys. 18, 681±684. Sanborn, E. N. and Beard, E. L. (1967) Sul®des of strontium, calcium, and magnesium in infrared-stimulated luminescence dosimetry. In: Proc. 1st Int. Conf. on Luminescence Dosimetry, pp. 183±191. Tochilin, E., Goldstein, N. and Miller, W. G. (1969) Beryllium oxide as a thermoluminescent dosimeter. Health Phys. 16, 1±7.


Walker, F. G., Colyott, L. E., Agersnap, Larsen N. and McKeever, S. W. S. (1996) The Wavelength Dependence of Light-Induced Fading of Thermoluminescence from a-Al2O3:C. Radiat. Meas. 26, 711±718. Wintle, A. G. (1997) Special Issue. Radiat. Meas. 27 (in press). Yoder, R. C. and Salasky, M. R. (1997) A dosimetry system based on delayed optically stimulated luminescence. Health Phys. 72, S18±S19.