Radiation Measuremtn~s, Vol. 23. Nos 213. pp. 519-522. 1994 Comricht (0 1994 Elsevier Science Ltd Rinted~iiGrarBritein.All rights resend 135&4487/94 s7.w + .oo A COMPACT DESIGN FOR MONOCHROMATIC OSL MEASUREMENTS IN THE WAVELENGTH RANGE 380-1020 NM L. J&TIER-JENSEN,N. R. J. POOLTON, F. W~UJMSENand H. CHRUIIANSEN Rise National Laboratory, DK-4000 Roskilde, Denmark Abstract-The development and performance of a compact module is described that allows for the monochromatic illumination of samples in the wavelength range 380-1020 nm, enabling the measurement of energy-resolved optically stimulated luminescence. The unit is designed to couple directly to the existing automated Riser TL/OSL dating apparatus, thus allowing for either routine scanning or more detailed [email protected] investigations. The high throughput efficiency of the unit means that the existing 75 W tungsten-halogen lamp can be directly used for such measurements on both quartz and feldspar samples. The design allows for rapid spectral scanning with a choice of resolution of anywhere between 10 and 80nm: stray light levels are less than 0.01%. The unit can equally be used for recording waveiengthresolved emission spectra, whether photo-excited or thermally stimulated; the capabilities of the system are demonstrated in the article. 1. INTRODUCTION 2. CONSTRUCTION THE USE of optically stimulated luminescence (OSL) in dating applications is now well established, and a number of different stimulation sources are now commonly used. These include the 514nm argon-ion laser emission (Rhodes, 1988), infrared emissions from light emitting diodes (BetterJensen et al., 1991) and broad band emitters such as incandescent or arc lamps, in conjunction with carefully selected optical filters (Spooner and Questiaux, 1990; Better-Jensen and Duller, 1992). Ideally, however, the spectral emission and excitation characteristics of quartz and feldspar materials prepared for dosimetric evaluation would be routinely scanned since, apart from giving valuable information on the mineralogy, it would also allow the possibility of choosing the most suitable energy windows in which to carry out the measurements. We show in this article that high quality excitation and emission spectra of quartz and feldspars can easily be obtained using a simple unit that directly attaches to the OSL unit of the automated Rise TL/OSL reader (BPrtter-Jensen and Duller, 1992) using as an excitation source the 75 W tungsten halogen lamp already incorporated. As described, the instrument is perhaps most closely related to the rapid scanning interference filter monochromator constructed by Bailiff et al. (1977). NEW RIS0 AND DESIGN OF THE MONOCHROMATOR By using a variable interference iilter as the wavelength dispersive element, very compact monochromator systems can be designed having high light throughputs and enabling very rapid spectral scanning. The instrument we present here can be used for a large range of different experiments either as a stand alone unit, or attached directly to the automated TL/OSL dating apparatus. The design is based around the use of two filters covering the spectral range of interest (the first usable in the range 380-75Onm and the second from 740-1020 nm). For UV/visible measurements, the Schott filter VERIL S-60 was used which gives spectral resolution of 1Onm across the entire range, transmission factors of around 35% and blocking to better than 0.01%. (The full characteristics are discussed in more detail below.) The infrared filter was specially designed and built for us by Barr Associates, and has a similarly high performance (spectral resolution of 9nm, transmission 60%). We have found that one advantage of splitting the spectral range is that UV/visible emissionspectra can be recorded at the same time as producing variable wavelength, monochromatic excitation, thus giving a complete luminescence excitation/emission spec trometer that plugs straight onto the automated Rise 519 L. BIFITER-JENSEN 520 reader. A demonstration of the capabilities is given in Section 3. In Fig. l(a), we show a detailed cross-section of the construction, demonstrating schematically in Fig. I(b) how it fits onto the existing dating apparatus; it adds only 4cm to the overall dimensions of the system, since all the focusing and collimating optics are placed in spaces already present. In the excitation mode, collimated light from the 75 W tungsten halogen lamp (or external source) is focused onto a slit, behind which the filter is scanned; the emerging monochromatic beam is then recollimated, and finally refocused onto the sample. Our construction allows the slit width to be continuously adjusted from 0 to IOmm, allowing flexibility in choice of spectral resolution down to 80 nm: it should be pointed out that with slit widths less than 1 mm, no increase in resolution is achieved, since the limiting factor then becomes the filter bandwidth (in our case, around 10 nm). The filter is mounted on a precision linear rail and is moved by means of a small stepper motor: this configuration allows for high accuracy in the positioning. In our system, the travel speed is restricted to between 0 and 30 rim/s (at 0.14 nm per step) for the visible system, or 50 rim/s (at 0.23 nm per step) with the infrared filter, giving, in each case, minimum scan times of around 15 s for the entire spectral range. In Fig. 2(a), we show the transmitted spectrum from a high pressure mercury arc source, a He-Ne laser and the output of a calibrated high resolution spectrometer set at 880nm. We use this diagram to highlight a number of points. Firstly, since the variation away from linearity of wavelength vs position INTERFERENCE SLIT FILTER I.R. 1 ‘SAMPLE EXTERNAL SOURCE DIODES Fw. 1. Detailed cross-section of the Rise monochromator (A) and a schematlc representation (B) of how it can be used on the dating apparatus in either detection (mono 1) or excitation (mono 2) mode. er al. HeNe 400 600 H.Res.Mon. 8&I 1000 b 00400 00 400 600 800 1000 600 800 1000 WAVELENGTH (nm) FIG. 2. Technical aspects of the monochromator: (a) resolution, linearity and stray tight rejection using a number of sources; (b) transmission factors of the linear filters; (c) demonstrating their use as an absorption spectrometer; (d) the power throughput of the system when coupled to a 75 W tungsten-halogen source. along the filters is very small (certainly within the spectral bandwidth), the wavelength setting can be directly measured by counting the number of steps advanced by the stepper motor (each step representing around 0.1 nm). Secondly, it will be noted that the spectral bandwidth (at 10nm) is nearly constant along both the filters. Thirdly, the laser line can be used to determine the stray light levels of the system. We have determined that at 450 nm, transmission of the 633 nm laser line is lower than 0.005%, although this increases to around 0.05% at 580 nm: similar figures are measured for stray light throughputs of the infrared filter. The transmission characteristics for the filters are shown in Fig. 2(b). Note that they are virtually spectrally neutral (unlike, for example, diffraction gratings), which makes correction to spectra for instrumental response straightforward. For the “visible” filter, transmission changes from 31 to 38% between 400 and 700 nm, respectively, and for the infrared system, from 50 to 60% in the range 750-980 nm. These figures are measured using f;! optics with two monochromators working in tandem, A MONOCHROMATOR the first acting as an absorption spectrometer and the second as the “sample”. Using a similar configuration, we show in Fig. 2(c) the transmission characteristics of the two broad band interference filters commonly used in OSL experiments. Finally, in Fig. 2(d), the power throughput of the system is shown, when linked to the 75 W quartz-tungsten-halogen lamp normally used for “broad band” excitation of OSL. We have installed a rear reflecting mirror behind the lamp, thereby increasing the illumination intensity by around 90%. and giving better filling of the monochromator slit. The power ranges from 0.01 to 1.1 mW/cm’/nm in the range 4QO-1000 run, following quite closely the blackbody curve for an emitter at 3400 K: the power transmittance is not entirely smooth due to the use of interference filters for heat rejection. Measurements were made with a calibrated pyrometer. Whilst the power throughput may appear low, we have never found that there is a shortage of excitation intensity, and the fall off in power below 5OOnm is in fact a great advantage, since the efficiency of exciting trapped charge increases very rapidly at the shorter wavelengths; the two effects therefore partly compensate, making repeated scans possible whilst only marginally affecting the trapped charge population. This effect is described in more detail in Section 3. 3. EXPERIMENTAL 521 FOR OSL MEASUREMENTS CAPABILI-ITES 3.1. Thermoluminescence emission spectra As mentioned in Section 2, the particular construction we have chosen means that the spectral range 380-700 nm can be scanned in around 15 s, taking a further 25 s to reset. This consequently limits the heating rates usable if the TL emission spectra are to be continuously recorded: for example, a ramp rate of l”C/s would give a temperature resolution of 15°C to the spectra with a repetition measurement every 40°C. Naturally, better resolution, etc. can be achieved either by slowing the ramp rate further, or employing a faster motor. However, analysis of the TL emission spectra previously published (Luff and Townsend, 1993) shows that in general (at least for silicate-type minerals), there are not usually rapid changes of spectrum in the high temperature regions. Useful thermoluminescence emission spectra can be taken with our system, as demonstrated in Fig. 3(a) discussed in Section 3.1. In this case, however, the difficulties are slightly easier to overcome, since the spectra are easily compensated if the time decay characteristics of the OSL are known. Further, the fall in intensity of the OSL signal can be partly compensated by scanning the luminescence spectrum from high to low wavelengths, thereby making use of the rise of PM tube detection efficiency as the signal decays. Erasing the rapidly decaying initial OSL signal prior to measurement can also simplify correction procedures. Figure 3(b) shows a typical OSL emission curve for the same sample as Section 3.1. 3.2.2. Time-stable PL. By illuminating many feldspar samples with deep-UV light, time-stable photoluminescence can often be stimulated. We have found the simplest method of exciting this type of signal is to use the 75 W quartz-tungsten-halogen lamp in conjunction with U-340 filters. Although the UV output is very low, the luminescence e!Bciency is often orders of magnitude higher than the OSL signal. An example of such a spectrum is shown in Fig. 3(c) for the same sample as used in Fig. 3(a): as will be noted, many more luminescence transitions are probed in this kind of measurement. 1 A 1 a ‘,! ,&lyI, , 400 so0 600 loo0 C 0:. I 400 8 I 600 . I 800 . 1 loo0 for a museum specimen of Na feldspar. 3.2. Photoluminescence emission spectra 3.2.1. Time-decaying OSL. Comparison of the TL and OSL emission spectra can be very important if comparison of the dosimetric information coming from the two techniques is to be made. Since the OSL is obviously time dependent, the problems of scan rate and repetition cycle are essentially the same as WAVELENGTH (nm) FIG. 3. Demonstration of the capabilities of the monochromator system as applied to a sample of high albitc: (a) TL emission spectrum; (b) OSL emission spectrum; (c) photoluminescence emission spectrum; (d) OSL excitation spectra. Detection (D) or excitation (E) wavelengths are as indicated. 522 L. BmER-JENSEN 3.3. Luminescence excitation spectra 3.3.1. Time-decaying OSL. The original work of Hutt et al. (1988) demonstrated how important it is to analyse the photoexcitation spectra of feldspar, and our own work (Poolton et al., 1994) shows that detailed knowledge of the excitation spectra of natural samples provides essential information on the mineralogy. The monochromator described in this article is ideally suited for producing such spectra. As the OSL signal decays under constant illumination, consideration of procedures for correcting the excitation spectra produced must necessarily be made. This is a major problem if high power excitation sources are used (e.g., tunable laser systems) especially in multi-mineral samples, since the bleaching efficiency as a function of intensity and wavelength becomes complex. In our system, however, we have been able to largely overcome these problems by using a low power (75 W) tungsten-halogen lamp as the excitation source. Firstly, the output powers are quite low (see Section 2) and the intensity falls off rapidly, just as the photon capture cross-section for the trapped charge is rising rapidly. As a consequence, during a typical wavelength-resolved OSL excitation scan, only around 10% of the trapped charge is evicted; in order to produce stimulation spectra of the sample, therefore, only the intensity spectrum of the exciting lamp is required, and is not a running analysis of how much charge is lost. A routine check of the validity of this method can be made by repeatedly scanning the stimulation spectrum; the method is valid as long as the excitation spectra remain the same from one cycle to the next. An example of the OSL stimulation spectrum from feldspar is shown in Fig. 3(d), where comparison can be made with the emission from the same sample (Fig. 3(a)-(c)). We have found that, in contrast to feldspar, the stimulation efficiency for quartz is exponential with energy (Bnrtter-Jensen er al., 1994). 3.3.2. Time-stable PL. Whilst the energy-resolved stimulation spectrum of the time-decaying OSL emission of the sample probed the energy levels of the “donor” type defect (within a donor-(conduction band)-acceptor recombination cycle), the energy levels of the acceptor type defect can be probed by scanning the stimulating wavelengths of the timestable, Stokes shifted photoluminescence. Whilst such measurements are possible with our system, they usually require highly photosensitive material. er al. 4. CHOICE OF EXCITATION WAVELENGTH AND INFLUENCE ON DOSlMETRIC EVALUATION Eventually, the main concern for dating is that the most suitable choice of stimulation wavelength is made in the OSL measurements. The instrument described here allows the entire stimulation spectrum to be scanned with very little loss of charge, so the dose-dependent growth curves can be easily measured for all stimulating wavelengths (400-1OOOnm). In practice, we have found very little difference across this range (Btiter-Jensen er al., 1994). Acknowlcdgemcnrs-We are most grateful to Mr C. Karpof Barr Associates Inc. for designing the infrared interference filters and to Dr 0. Johnsen of Copenhagen University for the loan of samples. This work is funded under CEC contract SCl-CT92-0800and the Institute of Earth Studies, Arhus University. REFERENCES BaililTl. K., Morris D. A. and Aitken M. J. (1977) A rapid scanning interference spectrometer: application to low level thermoluminescence emission. J. Phys. E: Sci. rnsrrWn.10, 1156-I 160. Butter-Jensen L., Ditlefsen C. and Mejdahl V. (1991) Combined OSL (infrared) and TL studies of feldspars. Nucl. Tracks Radiat. Meas. 18, 257-263. Batter-Jensen L. and Duller G. A. T. (1992) A new system for measuring OSL from quartz samples. Nucl. Tracks R&t. Meas. 20, 549-553. Better-Jensen L., Duller G. A. T. and Poolton N. R. J. (1994) Excitation and emission spectrometry of stimulated luminescence from quartz and feldspars. Rodiut. Meas.23.613-616. Hutt G.. Jaek I. and Tchonka J. (1988) Optical dating: K-feldspars optical response stimulation spectra. Quut. Sci. Rec. 7, 381-385. Luff B. J. and Townsend P. D. (1993) High sensitivity thermoluminescence spectrometer. Meas. Sci. Technol. 4, 65-71. Poolton N. R. J.. Wter-Jensen L., Ypma P. J. M. and Johnsen 0. (1994) Influence of crystal structure on the optically stimulated luminescence properties of feldspars. Radiat. Meos. 23, 551-554. Rhodes E. J. (1988) Methodological considerations in the optical dating of quartz. Qua. Sci. Rev. 7, 395-400. Spooncr N. and Questiaux D. G. (1990) Optical dating Achenheim beyond the Ecmian using green and infrared stimulation. In Synopses From A Workshop on Long and Short Range Limits in Luminescence Doting, pp. 97-103. Oxford, April 1989. RLAHA Occasional Publications No. 9.