Quaternary Geochronology Lu signal with dose, beyond the dose region where the conventionally measured fast component of the OSL signal is close to saturation. A deposition of dust that makes up the loess is quasi- continuous, occurring in both glacial and interglacial where this lithostratigraphy can be established. Luminescence dating has been applied to a small number were found to be in broad agreement with the chronology ARTICLE IN PRESS C3 Corresponding author. Institute of Earth Environment, Chinese based on the assignment of the MIS boundary ages. At a greater depth, independent age control is provided by the magnetic polarity reversal of the Brunhes/Matuyama 1871-1014/$-see front matter r 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.quageo.2006.05.020 Academy of Sciences, 10 Fenghui S. Road, High-Tech Zone, Xi?An 710075, China. Tel.: +862988323864; fax: +862988320456. E-mail address: ych.lu@263.net (Y.C. Lu). periods. The palaeosols formed in the interglacials are visually recognisable in the field and can be identified by their higher magnetic susceptibility as measured in the laboratory (e.g. Liu et al., 1985; An et al., 1991). The chronology for these archives has not been established using numerical dating methods. For the youngest part of the deposit, a lithostratigraphy has been established (An and Lu, 1984), identifying loess and soil units. In order to provide a chronology, the boundaries between the soils and loess units within the deposits of loess sections in China. Early thermoluminescence (TL) dating studies (e.g. Lu et al., 1987, 1988, 1999; Forman, 1991) were followed by several dating studies based on infra-red stimulated luminescence (IRSL) (e.g. Musson et al., 1994; Frechen, 1999). Recent comparison of the IRSL signals from Japanese loess deposits have shown them to be underestimated when compared with the optically stimu- lated luminescence (OSL) measurements made on quartz (Watanuki et al., 2005). Most recently, Lu et al. (accepted) obtained a chronology using a multiple-aliquot OSL procedure on quartz and their ages for the last 130ka application of luminescence dating further back in time. Using the protocol the recuperated OSL signal is separated and used to estimate the equivalent dose for fine-grained quartz in Chinese loess. Ages obtained using equivalent doses determined by the recuperated OSL dating method for 12 samples from the last interglacial?glacial loess?palaeosol sequence are in agreement with multiple aliquot OSL ages and previous stratigraphic information. Ages obtained for four Chinese loess samples from close to the Brunhes/Matuyama (B/M) boundary were also consistent with the expected age of 776ka. The recuperated OSL dating method, therefore, can go beyond the last interglacial and cover the total Brunhes epoch in Chinese loess. r 2006 Elsevier Ltd. All rights reserved. Keywords: Recuperated OSL dating; Fine-grained quartz; Chinese loess 1. Introduction Chinese loess provides an outstanding archive of past climatic change (Liu et al., 1985; An et al., 1991). The spanning the last 130ka have been linked to the marine oxygen isotope records. This allowed the ages for the marine isotope stage (MIS) boundaries calculated by Martinson et al. (1987) to be assigned to the loess sections multiple-aliquot, two-part, regenerative-dose protocol, has been developed as a recuperated OSL dating method to extend the Research Recuperated OSL dating of fine-grained X.L. Wang a , Y.C. a SKLLQG, Institute of Earth Environment, Chinese b SKLED, Institute of Geology, China Earthquake c Institute of Geography and Earth Sciences, Univers Received 12 October 2005; Available online Abstract The thermally transferred optically stimulated luminescence (OSL) separated into two components, the recuperated and the basic transferred 1 (2006) 89?100 paper quartz in Chinese loess a,b,C3 , A.G. Wintle c Academy of Sciences, Xi?An 710075, China Administration, Beijing 100029, China ity of Wales, Aberystwyth SY23 3DB, UK accepted 1 May 2006 24 July 2006 from fine-grained quartz extracted from Chinese loess can be OSL signal. The recuperated OSL signal continues to grow www.elsevier.com/locate/quageo ARTICLE IN PRESS Geochronolo (B/M) boundary, when the Earth?s magnetic field changed from reversed to normal polarity (e.g. Heller and Liu, 1982; Liu et al., 1985). The most recent age estimate for the B/M boundary is 775.671.9ka, as obtained from 40 Ar/ 39 Ar dating of lava flows on the island of Maui (Coe et al., 2004). In this paper, we present a new OSL dating technique. Our previous studies of the thermally transferred OSL signal (Wang et al., accepted) have suggested that it can be used to extend the range of dating using OSL measure- ments. In this paper the ages obtained using this signal for samples taken from one site in China will be compared with the ages obtained using a multiple-aliquot OSL procedure for quartz (Lu et al., accepted) and ages inferred from the lithostratigraphic record. Since this data set covers only the last glacial?interglacial cycle (i.e. the last 130ka), the new technique is also checked using samples from the B/M boundary. 2. Luochuan section?previous luminescence studies The loess?palaeosol sections at Luochuan (351 45 0 N, 1091 25 0 E) in the centre of the Chinese Loess Plateau have been studied extensively, with their records of climate change in the last 130ka being of particular interest (e.g. Heller and Liu, 1982; Liu et al., 1985). Previously, two luminescence dating methods have been applied to the loess at Luochuan. Forman (1991) used the TL emission from polymineral fine grains. Lu et al. (1999) obtained 12 TL or IRSL ages for the loess (L 1 ) above the top of the first major palaeosol (S 1 ). Neither study was able to obtain reliable luminescence ages over 84ka. The need for a chronology that extends beyond the last interglacial was highlighted in a recent review of Chinese loess and its role in under- standing the evolution of the East Asian monsoon (Huang et al., 2000). In a more recent study, loess from the Heimugou section at Luochuan has been used to test the sensitivity-corrected multiple-aliquot regenerative-dose (MAR) OSL dating protocol (Lu et al., accepted). Using this protocol on 33 fine-grained quartz samples taken from the upper 13m of the sedimentary record, down to the base of the last interglacial palaeosol (S 1 ), good agreement was found with the ages expected from the previously assigned chronology in which lithological units were linked to the climatic changes in the marine oxygen isotope record. 3. Sampling, sample preparation and dose rate determination For this project, samples were collected from two sections at Luochuan, Heimugou and Potou. Twelve samples identical to some of those used in the OSL study of Lu et al. (accepted) were extracted from the upper 13m of the Heimugou section (Table A1 in Appendix). These included three samples from the upper part of the X.L. Wang et al. / Quaternary90 Holocene soil (S 0 ), five samples from the last glacial loess (L 1 ), and four samples from the first major palaeosol (S 1 ); the latter is considered to have formed during the last interglacial corresponding to the whole of MIS 5. At the Potou section, the B/M boundary is recorded in the lowest part of the eighth loess unit (L 8 ) at a depth of 53.4m (Heller and Liu, 1982; Liu et al., 1985). Two samples were taken at 52.9 and 53.1m in the eighth loess (L 8 ) and two samples at 53.8 and 54.3m in the eighth palaeosol (S 8 ) (Table A1) in order to provide two samples above, and two samples below, the level of the B/M magnetic reversal. The preparation of feldspar-free, 4?11mm diameter, quartz grains from the bulk sediment has been described in detail elsewhere (Wang et al., 2006), together with the equipment used for the OSL measurements. Neutron activation analysis was used to measure the uranium and thorium concentrations of each sample, and the potassium content was determined by flame spectrophotometry. The a efficiency factor was measured for each sample by comparing the OSL signals regenerated by a- and b- irradiation after optical bleaching by a sunlamp (SOL 2) for 3min, as suggested for TL measurements by Zhou et al. (1995). The dose rate information for all samples is given in Table A1 in Appendix; the values are taken from the previous study (Lu et al., accepted), except those for the B/M boundary. 4. Thermally transferred OSL signals 4.1. Measurement of the thermally transferred OSL (TT-OSL) signal For optical stimulation with blue light-emitting diodes (LEDs, 47575nm), the OSL signal observed in the violet/ near-UV part of the spectrum undergoes a rapid decay in the first 5s of stimulation time. Examples of such decays for naturally irradiated aliquots are shown in Fig. 1(a) for samples IEE210 (MAR OSL age, 1.270.2ka) and IEE223 (MAR OSL age, 50.772.6ka) and in Fig. 1(b) for sample IEE425 (from the B/M boundary). After continuous stimulation for 270s at 1251C (Measurement 1 in Fig. 1(a) and Fig. 1(b)), the stimulation was switched off and each aliquot was heated to 2601C for 10s. The aliquot was then cooled to 1251C and the optical stimulation was re-started (Measurement 2 in Fig. 1(c and d)). This resulted in a new, rapidly decaying, OSL signal being superimposed on the slowly decaying component of the main OSL signals in Fig. 1; this is also shown on a linear ordinate scale as a function of time in the insets of Fig. 1(c and d). This TT-OSL signal has been studied in detail by Wang et al. (accepted); this signal contains the OSL signal to be used for dating in the current study. 4.2. Composition of the TT-OSL signal The TT-OSL signal measured after such a thermal treatment has been known since the earliest studies on the gy 1 (2006) 89?100 OSL signals of sediments (Huntley et al., 1985; Smith et al., 1986; Aitken and Smith, 1988). TT-OSL signals have been ARTICLE IN PRESS IEE210 IEE223 Geochronolo Measurement 1 10 100 1000 10000 100000 0 90 180 OSL intensity(cps) (a) 10000 100000 (b) X.L. Wang et al. / Quaternary shown to be made up of two types of OSL signals, termed recuperation and basic transfer (Aitken, 1998). In order for the recuperation effect to occur, electrons will have been transferred into a refuge trap from the OSL traps by light exposure (Aitken and Smith, 1988; Aitken, 1998). In this paper, we define OSL traps as those that give rise to the OSL signal on stimulation at 1251C. This transfer was occurring whilst other electrons were recom- bining at luminescence centres that give rise to the OSL signal. Then, thermal transfer of the electrons from the relatively thermally unstable refuge traps into the now empty OSL traps is caused by a subsequent thermal treatment. The next exposure to light gives rise to a new, smaller OSL signal?the recuperated OSL signal. This signal would be expected to be dose-dependent as it is linked to the electrons in the OSL traps prior to the first light exposure, i.e. those that had built up since the exposure to sunlight when the grains had been deposited. Thus the recuperated OSL signal will also be zeroed by light exposure and make it suitable for dating sediments. In contrast, the basic transferred OSL signal has no potential zeroing mechanism as it is derived from electrons 10 100 1000 0 90 180 Stimulation time1 (s) OSL intensity (cps) IEE Fig. 1. OSL decay curves of natural fine-grained quartz stimulated by blue LEDs IEE210 and sample IEE223, and (b) for sample IEE425 (c and d). After preheating (Measurement 2 with Stimulation time (2), and is also shown on a linear scale as and then in 10s intervals for the next 40s for every 90s measured. Measurement 2Preheating 0306090270 0 200 400 600 0306090 (c) 150 300 450 (d) gy 1 (2006) 89?100 91 trapped in light-insensitive traps prior to deposition of the grains (Rhodes, 1988; Aitken, 1998); it is thus unsuitable as a dating tool for detrital sediments. In addition, any signal growth with dose would be additional to that which results from electrons that come from the light-insensitive traps with a long lifetime, which are well populated before deposition and may also continue to accumulate electrons after deposition. Electrons in these traps are unaffected by laboratory light exposure, but a small fraction of them are transferred into the OSL traps by a sufficiently high thermal treatment in the laboratory (e.g. 2601C for 10s). Therefore the basic transferred OSL contribution should be excluded when dating detrital deposits. 4.3. Methods of separating the recuperated and basic transfer signals The TT-OSL signal can be separated into its component partsbyrepeatedlypreheating(holdingat2601Cfor10s) and optically stimulating (at 1251C for 90s) the same aliquot. This can be seen in Fig. 2(a) for an aliquot of sample IEE425 (from the B/M boundary) where the TT-OSL 0306090270 0 03 0 60 90 Stimulation time 2 (s) 425 for 270s (Measurement 1 with Stimulation time (1) for (a) sample discs at 2601C for 10s the TT-OSL signal is measured for 90s a function of time in the insets. Data are collected in 1s intervals for 50s intensity (obtained by integrating the first 5s of the signal and subtracting the average signal from the end of the 90s stimulation) is shown as a function of preheat/OSL cycle. The measured signals (L TTOSL ) are corrected for sensitivity change by measuring the OSL response (T TTOSL )toatest dose given after each TT-OSL measurement. The corrected TT-OSL decreases to a constant value of 0.02370.003 after 6cycles(Fig. 2(a)). Wang et al. (accepted) have applied the same procedure to a last interglacial sample IEE266 (MAR OSL age, 12678ka), and they also found no further reduction in the signal after 6 cycles (Fig. 5(a) of Wang et al., accepted). The constant signal remaining after 6 cycles was interpreted as the basic transfer signal, and the signal that disappears in the first 6 cycles was believed to be the recuperated OSL signal. It thus seems possible to separate the light-sensitive and dose-dependent recuperated OSL signal from the light- insensitive basic transferred OSL signal, though the above procedure (repeated preheating and stimulation) is not suited to the development of a rapid measurement procedure for dating purposes. Instead, we chose to investigate the use of a short thermal treatment above 2601Ctoseparatethe TT-OSL signal into its constituent parts. ARTICLE IN PRESS 0.25 0.2 0.15 L TTOSL X.L. Wang et al. / Quaternary Geochronolo92 0.1 0.05 0 04 10 0.25 0.2 0.15 0.1 0.05 0 200 250 300 350 400 Preheating and stimulation cycle Annealing(1 s) / Preheating (10 s) Corrected Corrected L TTOSL (a) (b) 268 300 ?C 360 ?C Fig. 2. Sensitivity-corrected thermally transferred OSL signal (L TTOSL / T TTOSL ) plotted (a) for repeated cycles of preheat (2601C for 10s) and blue stimulation (1251C for 90s) as a function of cycle andas a functionof (b) annealing (1s, &) or preheating (10s, K) temperature prior to the measurement. 4.4. Effects on the TT-OSL of heating for 1s Experiments were carried out on one of the samples from below the B/M boundary (sample IEE425). Twelve naturally irradiated aliquots were first preheated at 2601C for 10s to remove electrons from any thermally unstable traps. The aliquots were then exposed to blue light for 270s at 1251C to empty the OSL traps and cause electrons to be transferred into the refuge traps. Each aliquot was then heated to, and held for 1s at, a temperature between 200 and 4001C (in 201C intervals) in order to induce the thermally transferred signals. This short thermal treatment is referred to as ??annealing??. The resulting TT-OSL intensities (L TTOSL ) were measured using blue light stimulation for 90s at 1251C. In order to correct for sensitivity change, in much the same way as in the SAR protocol (Murray and Wintle, 2000), the OSL response (T TTOSL ) to a test dose of 7.8Gy was measured, but with a preheat of 20s at 2201C being used to remove the electrons from the thermally unstable traps. This correction proce- dure assumes that the OSL signal from the test dose is appropriate for monitoring the TT-OSL signal. This assumption is currently being tested. Since the TT-OSL and OSL signals are the direct result of electrons being optically ejected from the same traps, it would be expected that the electrons would recombine at the same lumines- cence centre. The relationship between the sensitivity-corrected TT- OSL intensity and the annealing temperature is shown in Fig. 2(b) (&). The 10-fold enhancement in the TT-OSL intensity as the annealing temperature is increased from 200 to 3201C is primarily due to the enhancement of the recuperated OSL signal as more electrons are thermally released from the refuge traps and captured at the OSL traps. As the annealing temperature is increased beyond 3201C, this effect is overtaken by the thermal ejection of the electrons from the OSL traps. Wang et al. (accepted) showed that the parameters (trap depth and frequency factor) controlling the thermal stability of the OSL traps into which electrons had been thermally transferred by heating for 10s at 2601C were identical (within experi- mental errors) to those for the main OSL signal. Given the thermal stability of the OSL traps, it would thus be expected that no signal would be left after holding for 1s at 360 or 3801C. However, a TT-OSL signal is seen (Fig. 2(b)); this is inferred to be the basic transfer signal. 4.5. Selection of preheat conditions for Part 1 of the dating protocol A preheat of 2601C for 10s was selected for Step 1-2 of the new dating protocol (Table 1, Part 1) in order to remove electrons from thermally unstable traps, whilst not touching the electrons in the OSL traps; these would be emptied at temperatures of 3201C and above (Fig. 2(b)). gy 1 (2006) 89?100 The same preheat condition was used in Step 1-4 to transfer the electrons from the refuge traps to the OSL ARTICLE IN PRESS TTOSL TTOSL BTOSL BTOSL Geochronolo traps as demonstrated in Fig. 1. This thermal treatment induces a TT-OSL signal equal to that after using a preheat of 2801C for 1s (Wang et al., accepted); however, for the dating protocol, thermal treatments of 10s are preferred to those for 1s because of better reproducibility. 4.6. Selection of preheat conditions for Part 2 of the dating protocol Following the measurements used to obtain the TT-OSL signal in Part 1 of the protocol, it is necessary to select an appropriate thermal treatment for Step 2-1. This is the step that would remove any remaining recuperated signal that is Table 1 Recuperated OSL dating protocol for each disc Step Experimental treatment Result Part 1. Detection of thermal-transferred OSL signal 1-1 Dose, D i ? 1-2 Preheating at 2601C for 10s ? 1-3 Blue stimulation at 1251C for 270s ? 1-4 Preheating at 2601C for 10s ? 1-5 Blue stimulation at 1251C for 90s L 1-6 Give test dose, D t ? 1-7 Preheating at 2201C for 20s ? 1-8 Blue stimulation at 1251C for 90s T Part 2. Detection of basic-transferred OSL signal 2-1 Annealing to 3001C for 10s ? 2-2 Blue stimulation at 1251C for 90s ? 2-3 Preheating at 2601C for 10s ? 2-4 Blue stimulation at 1251C for 90s L 2-5 Give test dose, D t ? 2-6 Preheating at 2201C for 20s ? 2-7 Blue stimulation at 1251C for 90s T X.L. Wang et al. / Quaternary not removed by Steps 1-5 to 1-8. To investigate the effects on the TT-OSL signal of heating for 10s, a second experiment was carried out on sample IEE425. Twelve naturally irradiated discs were treated as in Table 1, but in Step 2-1 the preheat temperature (held for 10s) was varied, holding at 2601C, or between 280 and 3601Cin101C steps or in 201C steps between 360 and 4001C. The TT-OSL signals were measured (Step 2-4) and normalised with the OSL response to a test dose (Step 2-7), as shown in Fig. 2(b) (K). As the 10s preheat temperature in Step 2-1 was increased from 260 to 3001C, the sensitivity-corrected TT-OSL intensity decreased to the level equivalent to the 1s anneal data point at 3601C(Fig. 2(b)); this value of the intensity has been inferred to be the basic transfer signal (Section 4.4). When preheating for 10s at 3201C or above, the signal is effectively zero. From Fig. 2(b), it can be seen that heating for 10s at 3001C (or 3101C) results in a corrected TT-OSL signal of 0.020 (Fig. 2(b)). This is the same signal level that was obtained for the TT-OSL signals for the 1s anneal at 3601C, and is thus also inferred to be the basic transfer signal. More importantly, it is the same (within experi- mental errors) as the level of 0.02370.003 obtained for repeated cycling (Fig. 2(a)). Thus 3001C for 10s was selected as the thermal treatment in Step 2-1 of the dating protocol (Table 1, Part 2) to remove the part of the recuperated OSL signal that survived from Step 1-4 in Table 1. For Step 2-3 the preheat was chosen to be 2601C for 10s to transfer the electrons from the refuge traps to the OSL traps (as in Step 1-4). 5. The recuperated OSL dating method The new dating protocol based on the recuperated OSL signal is given in Table 1, in which the preheat conditions have been selected on the basis of the experimental data of Note For natural samples, D i ? 0Gy Removing electrons in unstable TL traps Bleaching OSL signals Thermally inducing thermal-transferred OSL signals Detecting thermal-transferred OSL signals Monitoring the OSL production in quartz Removing electrons in unstable TL traps Measuring the test dose OSL response Thermally inducing remnant recuperated OSL signals Removing remnant recuperated OSL signals Thermally inducing basic-transferred OSL signals Measuring basic-transferred OSL intensity Monitoring the OSL production in quartz Removing electrons in unstable TL traps Measuring the test dose OSL response gy 1 (2006) 89?100 93 Wang et al. (accepted) and that given in Fig. 2 (Sections 4.5 and 4.6). First, five or more natural aliquots are measured using the protocol in Table 1 to obtain the intensity of natural L TTOSL and L BTOSL , respectively; the uncorrected L ReOSL intensity would be calculated as L ReOSL ? L TTOSL C0 L BTOSL . (1) For investigation of sensitivity changes during measure- ments and also for inter-aliquot normalization, two test doses (D t in Steps 1-6 and 2-5 in Table 1) were interjected after measuring the total thermal-transfer and basic transferred OSL signals. The corrected recuperated in- tensity, Corrected L ReOSL , was then calculated as Corrected L ReOSL ??L TTOSL =T TTOSL C138C0?L BTOSL =T BTOSL C138. (2) Each disc was then given a different regenerative dose (D i ) and all the steps in Table 1 were then repeated for each aliquot and the corrected recuperated OSL (L ReOSL ) signal for each regeneration dose was also calculated using Eq. (2). The recuperated OSL D e value for a particular sample was then determined by matching the corrected also demonstrate the large error that is attached to the D e value obtained for young samples using Eq. (2), as a result of the values of [L TTOSL /T TTOSL ] being not much larger than those for [L BTOSL /T BTOSL ]. In addition, the estimated recuperated signal intensities for these samples (L TTOSL and L BTOSL ) were small. For samples IEE209 and IEE210, with values of D e of 10.072.1 and 10.472.4Gy, respec- tively, the uncertainty is C2422%. 7.2. Samples from the last glacial/interglacial cycle (15?130ka) Nine samples, five from the L 1 loess and four from the underlying S 1 soil, were dated using the protocol in Table 1. Fig. 4(a?d) shows the raw data obtained for sample IEE219 taken from the L 1 loess at a depth of 3.5m. The measured thermally transferred and basic transferred OSL signals are shown as a function of regeneration dose in Fig. 4(e) and the dose?response curve for the recuperated ARTICLE IN PRESS 0.2 0.1 0 0 1000 2000 3000 4000 Dose (Gy) Corrected L (b) Fig. 3. Sensitivity-corrected (a) OSL (L i /T i ) and (b) recuperated OSL (L ReOSL ) intensity as a function of laboratory dose, with each data point derived from a single aliquot for young sample (IEE209) given radiation doses in addition to the small environmental dose of about 3Gy. Signal integration time was 5s. The dashed line in Fig. 3(b) is linear fitting of recuperated OSL intensity against higher doses beyond 1000Gy, for comparison with the linear trend in Fig. 6(a) for sample IEE424. Both used continuous irradiation at room temperature. Geochronolo natural L ReOSL intensity to the dose?response curve constructed using the corrected regenerated L ReOSL in- tensities. The method is thus a sensitivity-corrected multi- ple-aliquot regenerative-dose protocol, in which the 15 steps in Table 1 are repeated twice for each aliquot. It is possible that further repetition of these 15 steps could be used as the basis of a single aliquot protocol for recuperated OSL dating. It should be pointed out that the corrected L ReOSL intensities are obtained by taking the difference between two numbers. 6. Dose?response curves for a Holocene sample The sensitivity-corrected OSL response curve for sample IEE209, constructed using conventional measurements of the OSL signal, was obtained by giving radiation doses to different aliquots in addition to the small environmental dose of about 3Gy (Fig. 3(a)). The measurement procedure is that developed as the MAR protocol (Lu et al., accepted), with doses up to 3780Gy being added. The non-linear nature of the dose?response curve obtained for the OSL signals is demonstrated in Fig. 3(a). A dose?response curve for the L ReOSL signal (obtained using the procedure described in Section 5) is shown for aliquots from the same sample (Fig. 3(b)). For each aliquot a dose was given in addition to the small natural dose and the measurement protocol (Parts 1 and 2 of Table 1) was applied only once to obtain each data point. It can be seen that the recuperated OSL signal continues to grow for doses up to 3780Gy, the highest dose given here. 7. Application of the new protocol The dating protocol of Table 1 was applied to a number of samples. For the Holocene samples and those from the last glacial/interglacial cycle, 6 discs were used to construct the dose?response curves. For the samples from the B/M boundary, 8 discs were used. The sensitivity-corrected values of the natural recuperated OSL signal were projected onto the dose?response curve to obtain the D e value. The error term in the D e value is obtained using the deviation of each measured regeneration dose point from the curve that fits the data points and the standard error of the natural recuperated OSL intensity. 7.1. Holocene samples To test how well the recuperated OSL signal in Chinese loess is zeroed in nature, three Holocene samples (IEE208, 209 and 210 with fine-grained quartz OSL ages obtained using the MAR protocol (Lu et al., accepted)) of 1.170.1, 0.870.1 and 1.270.2ka, respectively) from the upper part of the Holocene soil (S 0 ) were dated using the recuperated OSL dating protocol of Table 1. Recuperated OSL dating gave ages of 5.470.4, 2.870.6 and 2.970.7ka, respectively X.L. Wang et al. / Quaternary94 (Table A1). Thus the recuperated OSL ages are over- estimated by a few thousand years. These three samples 0 30 20 10 0 1000 2000 3000 4000 0.4 0.3 ReOSL L i /T i (a) gy 1 (2006) 89?100 OSL signal calculated from Eq. (2) is given in Fig. 4(f), showing how the D e value of 12272.3Gy was obtained. This led to a recuperated OSL age of 3671.5ka for sample IEE219 (Table A1). For this sample, the D e value was obtained with very high precision (2%) but for all the older samples in this group, the average precision was 471%. This is the result of the small contribution of the basic transferred OSL signal relative to the TTOSL signal seen for all doses over 100Gy (Fig. 4(e)). The uncertainties in the corrected L ReOSL signal for each regeneration dose point (Fig. 4(f)) are mainly due to the low signal/noise ratio when the basic transferred OSL signal is measured; the signal is of a similar magnitude to the instrumental background. However, it must be mentioned that for sample IEE219, the standard error of natural recuperated OSL intensity is 1.03%, as obtained for the six measured discs. This indicates that there is excellent reproducibility of both the natural and the regenerated data points; thus it is to be expected that it will be possible to recover a D e value with high precision when using the protocol in Table 1. ARTICLE IN PRESS 0 300 600 900 1 100 0 1 0 400 800 1200 1 0 1 0 400 800 1200 1 0 300 600 900 1 0.06 0.04 0.05 0.03 10 100 25000 20000 15000 10000 5000 0 1 10 100 25000 20000 15000 10000 5000 0 1 10 100 10 100 20000 15000 10000 5000 10 100 25000 20000 15000 10000 5000 10 100 10 100 Stimulation time(s)Stimulation time(s) L TTOSL (cps) L TTOSL (cps) & L BTOSL /T BTOSL ReOSL L BTOSL (cps) L BTOSL (cps) T BTOSL (cps) T BTOSL (cps) BTOSLTTOSL T TTOSL (cps) T TTOSL (cps) 0 Gy 47 Gy 94 Gy 141 Gy 188 Gy 10(a) (c) (d) (b) sample the X.L. Wang et al. / Quaternary Geochronology 1 (2006) 89?100 95 0 0.02 0.00 50 100 150 200 Regeneration dose (Gy) L TTOSL /T TTOSL (e) Fig. 4. Recuperated OSL data (collected as stated in caption to Fig. 1) for (a) natural thermally transferred OSL signal (L TTOSL ) and (b) natural basic-transf (d) regenerated basic-transferred OSL signals; (e) Dose?response curves for OSL intensities and (f) the recuperated OSL signal (obtained using Eq. (2)) and to a test dose (7.8Gy) following their previous measurements. 0.02 0.00 0 50 100 150 200 Regeneration dose (Gy) Corrected L ReOSL D e =122?2.3 Gy (f) IEE219 (3.5m below the modern ground surface). Decay curves of erred OSL signal (L BTOSL ); (c) regenerated thermal-transferred OSL and sensitivity-corrected thermally transferred (&) and basic-transferred (?) the resulting D e determination. The insets (in a?d) are the OSL responses ARTICLE IN PRESS 30 S 0 0.12 0.06 0.00 0 1000 2000 3000 0.16 0.12 0.04 0.00 0 1000 2000 3000 0.16 0.12 D e = 1460?62 Gy D e = 2433?76 Gy (b) (a) Corr ected Corrected L ReOSL 0.08 Geochronolo All recuperated OSL ages from the last 15?130ka are given in Table A1.InFig. 5, the recuperated OSL ages are shown as a function of the OSL ages obtained by the sensitivity-corrected MAR protocol (Lu et al., accepted). The two sets of ages can be seen to be in agreement with each other within uncertainties. The recuperated OSL age estimates of five samples from the last glacial loess (L 1 ) agree with the SPECMAP ages (Martinson et al., 1987) for the transitions MIS 6/5 (129.873.1ka), MIS 5/4 (73.97 0 306090120 MAR OSL age (ka) MIS 2 / 1 L 1 S 1 Fig. 5. A comparison of the recuperated OSL and MAR OSL ages for loess and palaeosols of the past 130ka at Heimugou section, Luochuan. The 1:1 line is given and the Marine Isotope Stage (MIS) boundaries are shown. 120 90 60 ReOSL age (ka) MIS 6 / 5 MIS 5 / 4 X.L. Wang et al. / Quaternary96 2.6ka) and MIS 2/1 (12.173.1ka). Sample IEE228, with a recuperated OSL age of 7273.5ka, was taken immediately above the lithostratigraphic position of the MIS 5/4 transition. Sample IEE266, with a recuperated OSL age of 12176.1ka, was taken from just above the base of the palaeosol S 1 that is equivalent to the lithostratigraphic position of the MIS 6/5 transition. This is taken as validation of the use of recuperated OSL dating for age determination of Chinese loess deposited in the past 130ka. 8. Dose?response curves for a sample from the B/M boundary 8.1. Using continuous laboratory irradiation Fig. 6(a) shows the L ReOSL dose?response curve regen- erated for an older sample (IEE424 from a depth of 53.8m) that would be obtained in the course of application of the protocol in Table 1. The values of L ReOSL were obtained using Eq. (2). In the construction of the dose response curve, optical bleaching ahead of the regeneration dose in the second cycle of the protocol (Table 1, Part 1, Step 1) would have been achieved during Steps 1-3, 1-5 and 1-8 in Part 1 and Steps 2-2, 2-4 and 2-7 in Part 2 of the protocol 0.24 0.18 L ReOSL gy 1 (2006) 89?100 to measure the natural signal. For laboratory doses above 1000Gy, the L ReOSL appears to grow linearly with dose up to about 3000Gy, as was found when adding dose to the very young sample IEE209 (Fig. 3(b) dotted line). When the sensitivity-corrected natural L ReOSL intensity is pro- jected onto the dose?response curve (Fig. 6(a)), a D e value of 1460Gy is obtained, resulting in an age of only 474ka. This age is significantly underestimated when compared 0.08 0.04 0.00 0 1000 2000 3000 D e = 1373?114 Gy D e = 2371?66 Gy (c) Corrected L ReOSL Fig. 6. Corrected recuperated OSL intensity as a function of laboratory dose, for older sample (IEE424) that had been laboratory bleached and given doses with (a) continuous and (b) pulsed irradiation. (c) After recuperated OSL D e determination illustrated in (b), eight aliquots were separated into two groups, given a known dose of 2355 and 1413Gy, respectively, and then a dose recovery test was carried out. The ?natural? signal is shown projected onto the dose?response curves giving rise to values of the equivalent dose. ARTICLE IN PRESS Geochronolo with the known age for the B/M boundary (776ka) from which it is taken. 8.2. Using pulsed laboratory irradiation The severe age underestimation suggests that there is some difference in the recuperated OSL production measured following laboratory irradiation and that mea- sured following irradiation in the natural environment. The most significant difference in irradiation conditions is the rate at which the dose is delivered, with a natural dose rate of about 3Gy/ka and a laboratory dose rate of 0.157Gy/s. Valladas and Ferriera (1980) found that by increasing their laboratory g-dose rate from 0.01 to 10Gy/s, the TL response of their quartz could be increased by C2410%. Such behaviour has been explained as being due to competition for electrons from relatively thermally un- stable traps that are effectively kept empty during natural irradiation (Aitken, 1985). Dose rate dependence has also been suggested as a possible problem for OSL dating of quartz (Chawla et al., 1998; Huntley and Prescott, 2001). A theoretical explanation was proposed by Chen and Leung (2000) using a model with one trap and two recombination centres. Using a similar model with two traps and one recombination centre, Chen and Leung (2001) were able to show a dose-rate dependence of a calculated OSL integral. It has been suggested that giving irradiation in pulses might provide a better simulation of the trapping condi- tions in the natural environment (Bailey, 2004). Since the recuperated OSL ages are not underestimated when compared with conventional OSL ages of less than 60ka in the Heimugou loess section (Fig. 5), the pulsed- irradiation interval was fixed at C24150Gy; this dose level is reached before onset of acute non-linearity of the quartz OSL dose?response curve. Assuming that the recuperation effect is involved with the 110, 160 and 2201C TL peaks in quartz (Aitken and Smith, 1988; Spooner et al., 1988; Smith and Rhodes, 1994; Aitken, 1998), the sample disc was heated to 2401C after each pulsed-irradiation interval. Fig. 6(b) illustrates D e determination using this approach. Projecting the natural L ReOSL signal onto this dose? response curve (Fig. 6(b)) gives a larger value for the D e , which will be discussed in the next section. 9. Application of the new protocol to samples from the B/M boundary Using the protocol in Table 1 for sample IEE424, but giving the laboratory doses in 150Gy pulses separated by heating to 2401C, resulted in the dose?response curve for the L ReOSL signal shown in Fig. 6(b) (repeated in Fig. A1 for completeness). The measured data given in Fig. A1(a?d) look very similar to that for the much younger sample (IEE219) shown in Fig. 4. For sample IEE424, the standard error for the natural recuperated X.L. Wang et al. / Quaternary OSL intensity is 2.4% for the eight discs measured. For this sample, and the other three samples from the B/M boundary, the precision in the D e value was 471%, similar to that found for the samples from the sediment of the last glacial and interglacial. The D e value of 2433776Gy combined with a dose rate of 3.0870.12Gy/ka (Table A1) resulted in a recuperated OSL age of 790739ka for sample IEE424 (0.4 beneath the B/M boundary), an acceptable age for the B/M boundary. The four samples taken close to the B/M boundary gave an average age of 771715ka when using recuperated OSL dating, consistent with the recent 40 Ar/ 39 Ar age of 775.671.9ka recorded in lava flows (Coe et al., 2004). Further validation of using pulsed irradiation for old samples was obtained by giving known laboratory doses (2355Gy in 18 pulses and 1413Gy in 9 pulses) to aliquots previously used for D e determination of sample IEE424; the measured doses were 23717114 and 1373766Gy, respectively (Fig. 6(c)). The procedure is thus able to recover doses back to 2500Gy within experimental error; these dose recovery results indicate that it is suitable to use the test dose OSL response to allow for sensitivity changes in the recuperated OSL signal. It should be noted however that these doses were given after the aliquots had already experienced two cycles of the protocol. These results, and the ages for these four samples (Table A1), provide evidence that recuperated OSL dating offers a new method for dating the B/M boundary in Chinese loess. However, the apparent saturation of the corrected L ReOSL signal above 2500Gy (seen in Fig. 6(b and c)) suggests that this is close to the upper limit of the method. 10. Conclusions In this paper, we have developed and tested a multiple- aliquot, two-part, regenerative-dose protocol for the recuperated OSL dating of fine-grained quartz in Chinese loess. For samples from the Holocene soil (S 0 ), the procedure results in overestimation of the ages when compared with a more conventional optical dating method. In addition, because of the need to calculate the difference between the total TT-OSL signal and the basic transferred OSL signal, an error of C2420% in the D e value is obtained. Thus, this new method appears unsuitable for dating fine grained quartz of Holocene age. Good agreement of the recuperated OSL ages and conventional OSL ages (derived using the sensitivity- corrected MAR protocol, Lu et al. (accepted)) for samples from the last glacial loess (L 1 ) and the last interglacial palaeosol (S 1 ), covering the dating range from 15 to 130ka, is taken as validation for the new method in the late Quaternary. In addition the ages obtained with the new protocol are in broad agreement with the ages previously assigned for the lithostratigraphic boundaries identified in the Heimugou section at Luochuan. To obtain correct ages when applying the new protocol to samples from the B/M boundary at the Potou section at gy 1 (2006) 89?100 97 Luochuan, it was found necessary to give the laboratory doses in steps of 150Gy with intervening heating to 2401C. The recuperated OSL signals grow to high doses, with significant growth being shown for doses up to 2500Gy. In order to explain this extended dose response, and using the previousexplanationoftherecuperationeffect,itisnecessary to suggest that the production of conventional OSL signals is limited by the number of available luminescence centres. The number of trapped electrons continues to increase with dose to higher dose levels than that implied by the OSL dose?response curves; a portion of these trapped electrons can be sampled by the thermal transfer process. When these are measured, an adequate number of luminescence centres are available. Thus, we propose that recuperated OSL dating offers a new dating method that is able to go beyond the 130ka limit encountered in many detrital deposits when using conventional OSL protocols. It is now possible to obtain ages for samples with dose rates of C243Gy/kaforas far back as 0.8Ma. Acknowledgements We would like to dedicate this paper to the memory of Nicholas Shackleton who, over twenty years ago, encour- aged AGW to apply luminescence dating techniques to loess. We thank Prof. An Z.S., Prof. G.A.T. Duller and an anonymous referee for valuable comments on the manu- script. This work was funded by the National Basic Research Program of China (no. 2004CB720200) and the NSFC (no. 40523002). ARTICLE IN PRESS 0 500 1000 1500 2000 2500 1 10 100 L BTOSL (cps) L BTOSL (cps) 0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700 200 300 400 500 600 700 200 300 400 500 600 700 101 100 101 100 0 500 1000 1500 2000 2500 3000 1 10 100 L TTOSL (cps) L TTOSL (cps) 0 500 1000 1500 2000 2500 3000 1 101 100 3297 Gy 2826 Gy 2355 Gy 1884 Gy 1413 Gy 942 Gy 471 Gy 0 Gy 0 500 1000 1500 2000 2500 1 10 100 Stimulation time (s) Stimulation time (s) (a) (b) (d)(c) boundary). thermally-tran Appendix X.L. Wang et al. / Quaternary Geochronology 1 (2006) 89?10098 0.00 0.05 0.10 0.15 0.20 0 1000 2000 3000 4000 TTOSL BTOSL Regeneration dose (Gy) L TTOSL / T TTOSL & L BTOSL / T BTOSL (e) Fig. A1. Recuperated OSL dating of sample IEE424 (0.4m below the B/M thermally transferred and (b) basic-transferred OSL signals; (c) regenerated curves for (e) sensitivity-corrected thermally transferred (L TTOSL ,&) and basic-transfe corrected recuperated OSL intensities calculated using Eq. (2) (L ReOSL , J) obtained and d) are the OSL responses to a test dose (7.8Gy) after the preceding natural 0 1000 2000 3000 4000 Regeneration dose (Gy) 0.00 0.04 0.08 0.12 0.16 Corrected L ReOSL ReOSL D e = 2433?76 Gy (f) Decay curves (collected as stated in caption to Fig. 1) of (a) natural sferred and (d) basic-transferred OSL signals. Dose?response rred OSL (L BTOSL ,?) intensity with pulsed-irradiation and (f) The 0 100100 1 101 100 0 10 100 using pulsed irradiation and D e determination. The insets (in a, b, c and regenerated thermally and basic-transferred OSL measurements. ARTICLE IN PRESS Table A1 Data for MAR OSL and recuper ated OSL age de termina tion Lab. No Stratigraphy D epth (m) U (ppm) Th (ppm) K (%) Water content (%) Alpha coefficient Dose rate (Gy/ka) D e (Gy) Age (ka) MAR OSL Recuperated OSL MAR OSL R ecuperated OSL IEE208 S 0 0.1 2.19 7 0.08 11.50 7 0.25 1.86 15 7 2 0.040 3.48 7 0.11 3.6 7 0.2 18.7 7 1.2 1.0 7 0.1 5.4 7 0.4 IEE209 S 0 0.3 2.34 7 0.09 12.20 7 0.27 1.90 15 7 2 0.039 3.60 7 0.12 3.0 7 0.4 10 7 2.1 0.8 7 0.1 2.8 7 0.6 IEE210 S 0 0.5 2.46 7 0.10 13.10 7 0.29 1.91 18 7 2 0.039 3.57 7 0.12 4.3 7 0.6 10.4 7 2.4 1.2 7 0.2 2.9 7 0.7 IEE215 L 1 2.0 2.39 7 0.10 11.60 7 0.26 1.69 15 7 2 0.048 3.44 7 0.12 79.8 7 2.4 71.4 7 7.2 23.2 7 1.1 21 7 2.2 IEE219 L 1 3.5 2.31 7 0.09 12.20 7 0.27 2.03 20 7 2 0.035 3.35 7 0.12 110 7 1.5 122 7 2.3 32.8 7 1.2 36 7 1.5 IEE223 L 1 5.5 2.22 7 0.09 12.20 7 0.27 1.94 20 7 2 0.044 3.32 7 0.12 168 7 6.4 189 7 5.7 50.7 7 2.6 57 7 2.7 IEE272 L 1 8.0 2.43 7 0.13 12.01 7 0.26 1.84 15 7 2 0.045 3.50 7 0.13 214 7 15 239 7 7.3 61.2 7 4.8 68 7 3.3 IEE228 L 1 8.5 2.40 7 0.10 11.40 7 0.25 1.88 15 7 2 0.056 3.59 7 0.13 232 7 4.4 257 7 7.3 64.6 7 2.6 72 7 3.5 IEE232 S 1 9.5 2.55 7 0.10 13.80 7 0.30 2.01 25 7 2 0.052 3.35 7 0.16 270 7 12 305 7 18 80.6 7 5.3 91 7 7.0 IEE270 S 1 10.6 2.48 7 0.14 15.50 7 0.34 2.09 20 7 2 0.039 3.51 7 0.16 341 7 16 309 7 14 97.3 7 6.5 88 7 5.6 IEE268 S 1 11.5 2.72 7 0.15 13.45 7 0.30 2.07 20 7 2 0.054 3.48 7 0.16 354 7 11 372 7 16 102 7 5.8 107 7 6.7 IEE266 S 1 12.3 2.61 7 0.14 13.07 7 0.29 1.96 20 7 2 0.040 3.24 7 0.15 409 7 18 393 7 8 126 7 8.1 121 7 6.1 IEE421 L 8 52.9 2.62 7 0.13 13.00 7 0.29 1.83 23 7 2 0.034 3.08 7 0.12 ? 2382 7 88 ? 774 7 41 IEE422 L 8 53.1 2.46 7 0.11 13.30 7 0.29 1.83 25 7 2 0.041 3.04 7 0.12 ? 2296 7 86 ? 755 7 41 IEE424 S 8 53.8 2.50 7 0.11 14.00 7 0.31 1.90 25 7 2 0.034 3.08 7 0.12 ? 2433 7 76 ? 790 7 39 IEE425 S 8 54.3 2.67 7 0.13 13.80 7 0.32 1.83 25 7 2 0.040 3.06 7 0.12 ? 2339 7 118 ? 765 7 48 X.L. Wang et al. / Quaternary Geochronolo References Aitken, M.J., 1985. Thermoluminescence Dating. Academic Press, London. Aitken, M.J., 1998. An Introduction to Optical Dating. Oxford University Press, Oxford. Aitken, M.J., Smith, B.W., 1988. Optical dating: recuperation after bleaching. Quaternary Science Reviews 7, 387?393. An, Z.S., Kukla, G., Porter, S.C., Xiao, J.L., 1991. Magnetic susceptibility evidence of monsoon variation on the Loess Plateau of central China during the last 130,000 years. Quaternary Research 36, 29?36. An, Z.S., Lu, Y.C., 1984. A climatostratigraphic subdivision of Late Pleistocene strata named by Malan Formation in North China. Chinese Science Bulletin 29, 1239?1242. Bailey, R.M., 2004. Paper I?simulation of dose absorption in quartz over geological timescales and its implications for the precision and accuracy of optical dating. Radiation Measurements 38, 299?310. Chawla, S., Rao, T.K.G., Singhvi, A.K., 1998. Quartz thermolumines- cence: dose and dose-rate effects and their implications. Radiation Measurements 29, 53?63. Chen, R., Leung, P.L., 2000. A model for dose-rate dependence of thermoluminescence intensity. Journal of Physics D 33, 846?850. Chen, R., Leung, P.L., 2001. Nonlinear dose dependence and dose-rate dependence of optically stimulated luminescence and thermolumines- cence. Radiation Measurements 33, 475?481. Coe, R.S., Singer, B.S., Pringle, M.S., Zhao, X., 2004. Matyama-Brunhes reversal and Kamikatsura event on Maui: paleomagnetic directions, 40 Ar/ 39 Ar ages and implications. Earth and Planetary Science Letters 222, 667?684. Forman, S.L., 1991. Late Pleistocene chronology of loess deposition near Luochuan, China. Quaternary Research 36, 19?28. Frechen, M., 1999. Luminescence dating of loessic sediments from the Loess Plateau, China. Geologische Rundschau 87, 675?684. Heller, F., Liu, T.S., 1982. Magnetostratigraphical dating of loess deposits in China. Nature 300, 431?433. Huang, C.C., Pang, J.L., Zhao, J.P., 2000. Chinese loess and the evolution of the east Asian monsoon. Progress in Physical Geography 24, 75?96. Huntley, D.J., Godfrey-Smith, D.I., Thewalt, M.L.W., 1985. Optical dating of sediments. Nature 313, 105?107. Huntley, D.J., Prescott, J.R., 2001. Improved methodology and new thermoluminescence ages for the dune sequence in south-east South Australia. Quaternary Science Reviews 20, 687?699. Liu, T.S., et al., 1985. Loess and the Environment. China Ocean Press, Beijing. Lu, Y.C., Prescott, J.R., Robertson, G.B., Hutton, J.T., 1987. Thermo- luminescence dating of the Malan loess at Zhaitang, China. Geology 15, 603?605. Lu, Y.C., Wang, X.L., Wintle, A.G., Accepted. A new OSL chronology for dust accumulation in the last 130,000 years for the Chinese Loess Plateau. Quaternary Research. Lu, Y.C., Zhang, J.Z., Xie, J., 1988. Thermoluminescence dating of loess and paleosols from the Lantian section, Shaanxi Province, China. Quaternary Science Reviews 7, 245?250. Lu, Y.C., Zhao, H., Yin, G.M., Chen, J., Zhang, J.Z., 1999. Luminescence dating of loess?palaeosol sequences in the past about 100ka in North China. Bulletin of the National Museum of Japanese History 81, 209?220. Martinson, D.G., Pisias, N.G., Hays, J.D., et al., 1987. Age dating and the orbital theory of the ice ages: development of a high-resolution 0?300,000 year chronostratigraphy. Quaternary Research 27, 1?29. Murray, A.S., Wintle, A.G., 2000. Luminescence dating of quartz using an improved single-aliquot regenerative-dose protocol. Radiation Mea- surements 32, 57?73. Musson, F.M., Clarke, M.L., Wintle, A.G., 1994. Luminescence dating of gy 1 (2006) 89?100 99 loess from the Liujiapo section, central China. Quaternary Science Reviews 13, 407?410. Rhodes, E.J., 1988. Methodological considerations in the optical dating of quartz. Quaternary Science Reviews 7, 395?400. Smith, B.W., Aitken, M.J., Rhodes, E.J., Robinson, P.D., Geldard, D.M., 1986. Optical dating: methodological aspects. Radiation Protection Dosimetry 17, 229?233. Smith, B.W., Rhodes, E.J., 1994. Charge movements in quartz and their relevance to optical dating. Radiation Measurements 23, 329?334. Spooner, N.A., Prescott, J.R., Hutton, J.T., 1988. The effect of illumination wavelength on the bleaching of the thermoluminescence (TL) of quartz. Quaternary Science Reviews 7, 325?329. Valladas, G., Ferriera, J., 1980. On the dose-rate dependence of thermoluminescence response in quartz. Nuclear Instruments and Methods 175, 216?218. Wang, X.L., Lu, Y.C., Zhao, H., 2006. On the performances of the single- aliquot regenerative-dose (SAR) protocol for Chinese loess: fine quartz and polyminerals. Radiation Measurements 41, 1?8. Wang, X.L., Wintle, A.G., Lu, Y.C., Accepted. Thermally transferred luminescence in fine-grained quartz from Chinese loess: basic observa- tions. Radiation Measurements, doi:10.1016/j.radmeas.2006.01.001. Watanuki, T., Murray, A.S., Tsukamoto, S., 2005. Quartz and polymineral luminescence dating of Japanese loess over the last 0.6Ma: comparison with an independent chronology. Earth and Planetary Science Letters 240, 774?789. Zhou, L.P., Dodonov, A.E., Shackleton, N.J., 1995. Thermoluminescence dating of the Orkutsay loess section in the Tashkent region, Uzbekistan, Central Asia. Quaternary Science Reviews 14, 721?730. ARTICLE IN PRESS X.L. Wang et al. / Quaternary Geochronology 1 (2006) 89?100100