A comparison of natural- and laboratory-generated dose response curves for quartz optically stimulated luminescence signals from Chinese Loess

A comparison of natural- and laboratory-generated dose response curves for quartz optically stimulated luminescence signals from Chinese Loess

Radiation Measurements 47 (2012) 1045e1052 Contents lists available at SciVerse ScienceDirect Radiation Measurements journal homepage: www.elsevier...

713KB Sizes 0 Downloads 40 Views

Radiation Measurements 47 (2012) 1045e1052

Contents lists available at SciVerse ScienceDirect

Radiation Measurements journal homepage: www.elsevier.com/locate/radmeas

A comparison of natural- and laboratory-generated dose response curves for quartz optically stimulated luminescence signals from Chinese Loess M.S. Chapot a, *, H.M. Roberts a, G.A.T. Duller a, Z.P. Lai b a b

Institute of Geography and Earth Sciences, Aberystwyth University, Llandinam Bldg., Aberystwyth SY23 3DB, UK State Key Laboratory of Cryosphere Sciences, Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences, Lanzhou 730000, China

h i g h l i g h t s < Natural dose response curves can be made from profiles with independent age control. < Natural- and laboratory-generated dose response curves differ in shape. < Laboratory-generated dose response curves grow at higher doses than natural curves. < Component fitting and multiple aliquot protocols cannot resolve these differences. < These differences result in age underestimation above 150 Gy at Luochuan, China.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 9 February 2012 Received in revised form 14 August 2012 Accepted 1 September 2012

It has previously been observed that laboratory-generated quartz optically stimulated luminescence (OSL) signals from different samples have similar dose response curves (DRCs) after they are normalized using a test dose. It therefore seems likely that growth of the normalized signal due to natural irradiation of quartz may also follow a general dose response curve. The existence of such a curve is investigated by constructing a natural DRC from the test dose-normalized natural OSL signals of seven samples from the Luochuan section of the Chinese Loess Plateau. The same aliquots are then used to build single aliquot regenerative (SAR) DRCs, making it possible to compare the natural and laboratory constructed curves. Two main differences are observed. Firstly, the laboratory-generated DRCs are best fitted with double saturating exponential functions whereas the natural DRC is equally well fitted with a single saturating function. Secondly, in the laboratory-generated DRCs the normalized OSL signal continues to increase at high laboratory doses (>500 Gy), whereas no growth is seen at these doses in the equivalent natural DRC. These differences between natural- and laboratory-generated DRCs are still apparent even if data are manipulated to isolate the fast component, or if a sensitivity corrected multiple aliquot regenerative (SC-MAR) dose procedure is used. This suggests that the observed differences are not due to the influence of different components or inter-regenerative dose cycle sensitivity changes. The divergence between the natural- and laboratory-generated DRC means that the current maximum limit of quartz OSL dating at the Luochuan section is 150 Gy, as De estimates above this value are likely to be underestimations. Ó 2012 Elsevier Ltd. All rights reserved.

Keywords: Natural dose response curve OSL Loess Chinese loess plateau Luochuan Maximum age range

1. Introduction Investigations into standardized growth curves for the optically stimulated luminescence (OSL) signal from quartz have demonstrated that similar dose response curves (DRCs) can be built in the laboratory for samples collected in different regions and

* Corresponding author. Tel.: þ44 0 1970 622606. E-mail addresses: [email protected], [email protected] (M.S. Chapot). 1350-4487/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.radmeas.2012.09.001

sedimentary environments when aliquots are normalized with a sensitivity correcting test dose (Roberts and Duller, 2004). This idea has been applied in the use of common growth curves, based on similar DRCs from samples within the same study area, for determining equivalent doses for multi-grain aliquots (Lai et al., 2007; Lai, 2006; Stevens et al., 2007). The similarity in average behaviour observed between laboratory-constructed dose response curves from different samples opens up the possibility of the existence of a natural dose response curve, where growth of the normalized OSL signal from quartz due to natural irradiation over geological time follows a general dose response curve. If such

1046

M.S. Chapot et al. / Radiation Measurements 47 (2012) 1045e1052

a natural DRC exists, it would enable unprecedented comparison between OSL signal growth with radiation exposure in the laboratory and in nature, thereby enabling validation of the basic assumption of OSL dating; namely, that OSL signal growth due to beta irradiation in the laboratory is the same as that due to a lower intensity, mixed radiation field in nature. Furthermore, comparison of laboratory and natural dose response curves could provide a means of testing new techniques and protocols as luminescence techniques advance. In order to create a natural DRC, it is necessary to find a field site to act as a natural laboratory, i.e. where sediment has been accumulating for at least as long as it takes the environmental dose rate to saturate the quartz OSL signal. For multiple grain aliquots, it is necessary for all grains throughout the deposit to have had sufficient sunlight exposure to fully deplete the OSL signal before burial. It would be advantageous for the sediment to be homogenous and have similar source material so that luminescence sensitivity and environmental dose rates vary minimally throughout the deposit. Finally, it is essential that there exist some framework of independent age control that can be used to estimate the expected equivalent dose for each sample collected. One location that meets all of these criteria is the Chinese Loess Plateau. Wind-blown dust has accumulated in central China almost continuously for the past 2.6 million years (Liu, 1985), at least an order of magnitude longer than can be accurately dated with OSL in areas of reasonably high environmental dose rate. The dust originates from the deserts and desert margins north of the Plateau and becomes mixed and bleached during transport (Ding et al., 2002). Palaeosols formed in the loess deposits have been correlated with oxygen isotope records from marine cores and patterns of orbital cyclicity (Ding et al., 2002; Liu, 1985; Liu and Chang, 1964) providing a framework of independent age control that is visible in the field throughout the Chinese Loess Plateau. Some studies have demonstrated that quartz OSL from different loess units within a single site have reproducible DRCs (Stevens et al., 2007) and others (e.g. Lai, 2006; Roberts and Duller, 2004) have suggested that the growth curves are reproducible even between sites. The Luochuan section near the centre of the plateau has often been regarded as a type section (Liu, 1985). The stratigraphic profile at this section is nearly 130 m thick and fully exposed from the modern agricultural soil to the 2.6 Ma loess-red clay boundary (Kukla and An, 1989). Previous luminescence dating studies at this section have not only investigated the palaeoclimatic record, but also tested many new protocols against the framework of independent age control provided by the loess-palaeosol sequences. Several of the protocols tested have been successful for the uppermost loess (Forman, 1991; Lai, 2009; Lai et al., 2006; Lu et al., 2007; Wang et al., 2006) but others have reported age underestimations and other complications (Buylaert et al., 2007; Lai, 2009). Buylaert et al. (2007) observed age underestimation at the marine isotope stage (MIS) 4/5 boundary and therefore suggested that the maximum age limit of quartz OSL at Luochuan is about 40e50 ka, which corresponds to a maximum equivalent dose of about 120e 150 Gy. Lai (2009) also reported age underestimation compared to the loess-palaeosol sequences, his DRCs were best fitted with saturating exponential plus linear functions and the additional linear component was not reliable in producing accurate equivalent dose estimates above 230 Gy. This paper investigates the existence of a natural dose response curve at the Luochuan loess section and compares it to laboratory constructed dose response curves from the same material in order to test the similarity between the natural and laboratoryconstructed curves. The paper then goes on to investigate the maximum limit of OSL dating at this site.

2. Equipment and sample preparation Seven samples from the Luochuan section were used in this study (Fig. 1). All of the samples were collected from loess units in order to avoid post-depositional mixing and changes in environmental dose rate due to soil formation processes (Bateman et al., 2003). Six of the samples were taken within 60 cm of a palaeosol-loess boundary in order to minimize the distance from tie points with the framework of independent age control. The other sample was collected 2.0 m into the L1 loess and provides additional data in the conventional OSL age range. The samples were collected as small blocks wrapped in black plastic. Once they were in the darkroom conditions of the laboratory, exteriors of the blocks were removed and separated for dose rate calculations. The remaining material was treated with 10% volume to volume dilution of concentrated HCl and 20 vols H2O2 until no continued reaction could be identified. They were sieved to 35e63 mm, treated with H2SiF6 for 14 days to remove feldspar, and subsequently resieved as a further quartz purification step. Environmental dose rates were measured using the material that had been removed from the exterior of the blocks. This material was pulverised and homogenized before being measured using thick source alpha and beta counting. The modern water content was measured for each sample and gave an average value of 5  3%. As this value is likely to be lower than the average water content during burial because of desiccation from cliff-face exposure, the average burial water content was assumed to be 10  5%

Fig. 1. Pedostratigraphy of the uppermost 45 m of the Luochuan loess section demonstrating the location of OSL samples used for this study taken from loess units L1eL6. Paleosols are shown in black and unshaded regions are loess.

M.S. Chapot et al. / Radiation Measurements 47 (2012) 1045e1052

for all samples. The a-value was taken as 0.035  0.003 for quartz extracted from Chinese loess (Lai et al., 2008). The environmental dose rates for 35e63 mm grains of quartz varied from 3.15 to 3.43 Gy/ka (Table 1). Following sample preparation, luminescence measurements were performed on a Risø TL-DA-20 reader incorporating blue LEDs emitting at 470 nm delivering 50 mW/cm2 (Bøtter-Jensen et al., 2003). Samples were held at 125  C during optical stimulation for 50 s duration and the luminescence was recorded using an EMI9635QA photomultiplier tube equipped with two U-340 filters. A strontium/yttrium beta source with a dose rate of 0.095 Gy/s was used for laboratory irradiation and a preheat of 260  C for 10 s was applied to all samples before OSL stimulation for both the natural and test dose. All of the measurements used the signal integrated from 0 to 0.8 s (8 channels) minus a background integrated from 0.9 to 2.9 s (20 channels). It is important to ensure that the OSL signals for this study originate from quartz and not from feldspar, which may cause age underestimation due to anomalous fading. Previous studies have investigated methods of isolating the quartz signal both with room temperature infrared light stimulation prior to the blue light stimulation (double SAR (Banerjee et al., 2001), as applied to Chinese loess by Roberts and Wintle (2001)) and chemical treatment with H2SiF6 (Lai and Bruckner, 2008; Roberts, 2007). The results of these studies show that infrared stimulation does not completely remove the feldspar signal that can be stimulated with blue light and the double SAR protocol can still result in age underestimation in samples that have not been chemically treated. In order to isolate the quartz signal from samples in this study, both precautions were undertaken; the samples were subjected to 14 days H2SiF6 treatment, and a 40 s stimulation with infrared diodes at 20  C was also applied prior to each OSL measurement. An inverse OSL IR depletion ratio test (Duller, 2003) was added at the end of each sequence whereby the luminescence signal of a repeated radiation dose was measured without prior IR stimulation with the intent that measurements of any aliquot that were not within 10% of unity be discarded. However, no aliquot failed this criterion. Furthermore, De values measured with only blue OSL stimulation for all seven samples are consistent with post-IR blue OSL De values with a ratio of 0.98  0.03 (n ¼ 21) suggesting that all of the signals measured originate from quartz and the precautionary infrared stimulation had no impact on the results. 3. Investigating the existence of a natural dose response curve The environmental dose rate calculations were used to convert the framework of independent age control provided by the loess stratigraphy into a framework of expected equivalent doses (Table 1) using the following procedure. First, the expected age for each sample was determined by taking the age of the nearest

1047

palaeosol/loess transition from the Chiloparts record (Ding et al., 2002); this chronology was created by correlating grain size records from five loess sections, variations in the Earth’s obliquity and precession, and palaeomagnetic reversals. To determine the depositional age of each sample, a depth correction factor was added to account for the time between the palaeosol/loess boundary and the deposition of the sample. This factor was calculated by assuming a constant dust accumulation rate within the loess unit. Expected equivalent doses were determined by multiplying the expected age of the sample by its dose rate (Table 1). The error on the expected equivalent doses was estimated to be 10% for the six samples that were within 60 cm of a palaeosol/loess transition, and 15% for sample LC9, which was taken 2 m from the palaeosol/loess transition. To construct a natural dose response curve, the natural luminescence response (Ln) normalized to the response (Tn) to a test dose (25.4 Gy) was measured for six aliquots of each sample and plotted against the expected equivalent dose (Fig. 2). The results follow the expected shape of a quartz dose response curve but with large scatter between aliquots at high doses. The data are equally well described by either a single or a double saturating exponential function with adjusted R2 values (adjusted for the number of coefficients) of 0.84. When fitted with a single saturating exponential, this natural dose response curve has a D0 value of 98  10 Gy, which is similar to values summarised by Wintle and Murray (2006). However, the double saturating exponential D0 values of 35  46 and 181  96 Gy are lower than those reported by other authors based on laboratory irradiation to construct DRCs (Lowick et al., 2010; Murray et al., 2007; Pawley et al., 2010; TimarGabor et al., in press). 4. Laboratory dose response curves 4.1. SAR dose response curves The single aliquot regenerative (SAR) dose protocol for quartz OSL has been well studied (Murray and Wintle, 2003; Wintle and Murray, 2006) and is commonly used for estimating equivalent dose. In this protocol, the aliquot for which the natural luminescence signal is measured is then given a sequence of irradiation doses in the laboratory in order to construct an aliquot-specific dose response curve. In this study, such response curves were constructed for each aliquot that had been used to define the natural DRC (Fig. 2). The OSL measurements for regeneration doses (Lx) were performed under the same measurement conditions as the natural signals with regeneration doses of 0, 30, 60, 120, 240, 450, 750, and 1300 Gy, each followed by a measurement of the signal (Tx) induced by a test dose of 25.4 Gy. An additional zero and recycling regeneration dose was added to the end of the sequence for internal checks on the procedure and all of the aliquots demonstrated negligible recuperation and were able to recycle within 10% of unity.

Table 1 Sample details, including depth below palaeosol/loess boundaries. Ages for the boundaries are from correlation of the palaeosol/loess boundaries with marine isotope stages (Ding et al., 2002). The expected equivalent dose (De) for each sample was calculated by multiplying the environmental dose rate by the expected age. Sample

Depth below boundary (m)

Boundary age (ka)

Depth correction factor (ka)

Expected age (ka)

LC3 LC9 PT1 PT2 PT3 PT4 PT5

0.20 2.00 0.30 0.35 0.40 0.60 0.30

11 11 128 245 336 412 621

2 16 4 7 3 10 3

13 27 132 252 339 422 624

a

Dose rate for LC3 is unavailable and assumed to be the average of the other samples with 15% error.

      

1 4 13 25 35 42 62

Dose rate (Gy/ka) 3.25 3.20 3.18 3.15 3.43 3.17 3.31

      

0.5a 0.2 0.2 0.2 0.2 0.2 0.2

Expected De (Gy) 42.2 86.4 420 794 1163 1338 2065

      

8 14 50 94 135 158 241

1048

M.S. Chapot et al. / Radiation Measurements 47 (2012) 1045e1052

curves (Kim et al., 2009; Lai, 2009; Lowick et al., 2010). The data shown in Fig. 3 suggests that the continued growth only occurs in the laboratory, not in nature. 4.2. Fast-component isolated SAR dose response curve

Fig. 2. Six test-dose normalized natural optically stimulated luminescence signals (Ln/ Tn) for each of the seven samples plotted against its expected equivalent dose value (De). The data are fitted with both single and double saturating exponential curves, both of which have adjusted R2 values of 0.84.

All of the SAR DRCs are best fitted by double saturating exponentials with R2 values of 0.999 and higher (Fig. 3). If all of the data from the different aliquots are combined to construct a single dose response curve, it has an R2 value of 0.958 and D0 values of 49.6  8 and 383  43 Gy. These values are different to those obtained for the natural DRC (section 3) but are similar to the values of 44 and 450 Gy reported by Murray et al. (2007), the values of 51 and 320 Gy published by Pawley et al. (2010) and the values of 55  3 and 586  3 Gy obtained for coarse grains (63e90 mm) by TimarGabor et al. (in press). Unlike the natural DRC, laboratoryconstructed SAR DRCs cannot be well described by a single saturating exponential and continue to grow above the saturated Ln/Tn value observed in nature (a value of w3.85; Fig. 2). Continued growth of laboratory-constructed DRCs beyond that of a single saturating exponential function has been previously reported by many authors (Kim et al., 2009; Lai, 2009; Lowick et al., 2010; Murray et al., 2008; Pawley et al., 2010; Roberts and Duller, 2004) but concerns have been raised over the reliability of such growth

Fig. 3. A comparison of the natural DRC from Fig. 2 with single aliquot regenerative (SAR) dose response curves (DRCs) for each of the aliquots used to construct the natural DRC.

One explanation for the continued growth observed in the SAR DRCs (Fig. 3) is that the signal is complicated by contributions from more than one type of trap. It is known that the OSL signal from quartz is made up of a number of components which vary in thermal stability and saturation dose (Singarayer and Bailey, 2003). If the relative contributions of these components changes with dose, this may have an impact on the shape of the doseeresponse curve. This was investigated by Lowick et al. (2010) who component fitted CW-OSL decay curves and plotted the percentage contribution of each component to the total OSL as a function of stimulation time; they found no significant change with added dose. However, all of their measurements were based on the response to laboratory irradiation. In this study, we component fitted linearly modulated (LM) OSL measurements for both the natural DRC and the SAR DRCs to investigate whether isolation of the fast component would resolve or reduce the differences between the laboratory and natural DRCs. Component fitting of the LM-OSL decay curves was performed using commercial software SigmaPlotÔ (ver. 11.0) which employs the MarquardteLevenberg algorithm for linear and non-linear fitting. Each curve was fitted to the following equation: t2 t2 t2 t t t LðtÞ ¼ n1 b1 eb1 2P þ n2 b2 eb2 2P þ . þ nN bN ebN 2P P P P

(1)

Where L(t) is the luminescence intensity as a function of time (t), P is the duration of the luminescence measurement, nN is the number of trapped electrons in component N, bN is the detrapping probability of component N, which is proportional to the photoionisation cross-section (s), and I0 is the maximum stimulation light intensity, with b ¼ sI0. As the number of components has been shown to vary between samples (Jain et al., 2003; Singarayer and Bailey, 2003), all of the curves were fitted with 3, 4, and 5 components. The starting nN and bN values were 10,000 and 1.0 but were allowed to fluctuate throughout the iteration process. The maximum iterations, step size, and tolerance were 105, 102, and 10500 as recommended by Choi et al. (2006) and the fitting algorithm was weighted inversely by intensity. The number of components present was determined as the highest number of components before b values became redundant. Ultimately, all of the samples, aliquots and measurements were fitted with 4 components. After the initial curve fitting where starting bN values were allowed to fluctuate, the measurements were re-fitted with a set b1 value. This b1 value was the average b1 value of all measurements being used to construct a single dose response curve. For construction of the natural DRC, this consisted of 6 Ln and Tn measurements for each of the 7 samples, while for each SAR DRC this consisted of all the measurements recorded from a given aliquot. After component fitting all the measurements with 4 components and a set b1 value, b2 values were set in a similar manner and the component fitting algorithm was performed again. This process was repeated until b values for all the components were set. All nN values were allowed to fluctuate throughout the entire component fitting process. The residuals of the fitting procedure for a typical natural OSL decay curve are given in Fig. 4b along with a representation of the measured and fitted signal (Fig. 4a). For these samples the fast component is dominant, and makes up 95% of the net OSL signal. The photoionization cross-sections of the four components of the decay curves used to build the natural DRC are 2.98  0.19  1017 cm2,

M.S. Chapot et al. / Radiation Measurements 47 (2012) 1045e1052

1049

Fig. 4. a) LM-OSL signal (Ln) measured for one aliquot of LC3, along with the results of fitting 4 components. b) Residuals for the fitted function shown in (a) against measurement time.

6.00  0.56  1018 cm2, 4.04  0.23  1019 cm2, and 2.86  0.13  1020 cm2. The value for the fast component agrees well with the value published by Lowick et al. (2010) but is slightly higher than values published previously by Jain et al. (2003) and Singarayer and Bailey (2003). Nevertheless, the ratio of the medium and fast photoionization cross-sections of 0.23  0.02 is in agreement with the ratio of 0.24  0.03 published by these authors, suggesting that the components isolated in this study are the same components that were identified and investigated in previous studies. Isolation of the fast component reduces some of the scatter in the natural DRC (Fig. 5a) and less variability is seen in the form of the SAR DRCs (Fig. 5b) but the same relationships between natural and SAR DRCs (Fig. 5a and b) are observed as were seen using the net integrated signals (Fig. 3). The fast component isolated natural DRC can be equally well described by both single and double saturating exponential functions while the fast component isolated SAR DRCs are fitted significantly better with double saturating exponential functions. Additionally, Lx/Tx ratios of regeneration doses increase to values higher than the saturation Ln/Tn level of the natural DRC (Fig. 5b). Therefore it is concluded that the influence of different components is not the cause of the observed differences in DRC shape between the natural and SAR datasets. 4.3. MAR dose response curves If the difference in shape and saturation level between the natural and laboratory dose response curves is not due to contributions from more than one component, it may be the result of sensitivity change that was not accounted for by test dose sensitivity monitoring. SAR involves multiple regeneration cycles with sensitivity change between cycles being corrected for through the use of a test dose. To minimize cumulative sensitivity change, a sensitivity-corrected multiple aliquot regenerative (SC-MAR) dose OSL measurement protocol was used to generate a DRC. This protocol measures the OSL of each aliquot only twice, once for

Fig. 5. a) The natural DRC as constructed from fast-component isolated measurements of Ln/Tn values. b) A comparison of the natural DRC from (a) with fast-component isolated single aliquot regenerative (SAR) dose response curves (DRCs) for each of the aliquots used to construct the fast component isolated natural DRC.

a natural or a regeneration dose, and again for a test dose to measure the aliquot’s sensitivity. By subsequently plotting the Lx/Tx values against the given dose, the OSL signals of the different aliquots (Lx) are normalized by their test dose responses (Tx). SC-MAR DRCs were constructed for each of the samples used in this study using the same regeneration doses and test dose (25.4 Gy) as employed to construct the SAR DRCs. Every regeneration dose was measured using six aliquots for each sample. In order to remove the natural luminescence signal prior to irradiation, the aliquots were exposed to light for 3 min in a SOL2 simulator. Measurements on replicate aliquots from different samples that had received this bleaching treatment showed that the residual dose was 0.058  0.007 Gy, demonstrating the efficacy of this treatment in resetting the fast OSL signal. The preheats, test doses, and all other measurement conditions were the same as used previously for construction of the natural and SAR DRCs so that any variation observed in the resultant DRCs could be attributed to differences in measurement protocols rather than measurement conditions. When SC-MAR DRCs for all of the samples are compared with the natural DRC (Fig. 6), we observe the same relationships that were seen using both the net-integrated and fast-isolated SAR

1050

M.S. Chapot et al. / Radiation Measurements 47 (2012) 1045e1052

Fig. 6. A comparison of the natural DRC (Fig. 2) with sensitivity corrected multiple aliquot regenerative (SC-MAR) dose response curves (DRCs) for each of the samples used to construct the natural DRC.

DRCs. The Lx/Tx values obtained for high doses are similar to those from the SAR analyses, though there is less variability in the SCMAR DRCs which is likely to be due to averaging effects. This suggests that sensitivity change between regeneration dose cycles is not responsible for the differences in shape and saturation level of natural- and laboratory-created DRCs. However, comparisons of SC-MAR and SAR DRCs for individual samples demonstrate inter-sample variability potentially related to sample age. For samples LC3 (Fig. 7a) and LC9 (unit L1; Fig. 1), their SC-MAR DRCs saturate at lower Lx/Tx values than their SAR DRCs,

and the SC-MAR DRCs are in much closer agreement with the natural DRC. In contrast, samples PT1, PT2, and PT3 (units L2-L4; Fig. 1) exhibit no systematic difference between their SC-MAR and SAR DRCs (Fig. 7b) and samples PT4 and PT5 (units L5 and L6; Fig. 1) have SAR DRCs that grow to lower Lx/Tx values than their SC-MAR DRCs (Fig. 7c). For these samples the SAR DRCs are in closer agreement with the natural DRC. As De values can be obtained using the same Ln/Tn values on both the SC-MAR and SAR DRCs, curves that grow to higher Lx/Tx values result in lower De values. In previous comparisons between SC-MAR and SAR protocols, Lu et al. (2007) and Lowick and Preusser (2011) also had conflicting findings. Lu et al. (2007) found differences in De values between SCMAR and SAR protocols with SAR De values above 120 Gy underestimating the expected age by up to 13%. However, Lowick and Preusser (2011) demonstrated good agreement between SAR and SC-MAR DRCs and De values up to 400 Gy. Lu et al. (2007) hypothesized that the observed differences between the protocols originated from incomplete bleaching of regenerative doses by blue light stimulation at the end of each SAR cycle, but Lowick and Preusser (2011) suggested that the differences observed by Lu et al. (2007) were an effect of over-bleaching by the SOL2 simulator used for the SC-MAR procedure. In our study the same measurement conditions were used for all samples, but we see sample-specific differences between SAR and SC-MAR, suggesting that these differences reflect the age of the sample or quartz variability. 5. Comparing measured and expected De values In addition to comparing natural and laboratory constructed DRC shapes, one can also compare expected De values (Table 1) with those measured using SAR, fast component isolated SAR, and SC-

Fig. 7. Comparison of the single aliquot regenerative (SAR) dose response curves (DRCs) and sensitivity corrected multiple aliquot regenerative (SC-MAR) DRC for (a) sample LC3, (b) sample PT3, and (c) sample PT5.

M.S. Chapot et al. / Radiation Measurements 47 (2012) 1045e1052 Table 2 Comparison of expected equivalent dose values obtained in Table 1 with the De values measured using different procedures described in the text. Sample

Expected De (Gy)

LC3 LC9 PT1 PT2 PT3 PT4 PT5

42.2 86.4 420 794 1163 1338 2065

      

8 14 50 94 135 158 241

SAR De (Gy) 50 91 281 356 397 417 417

      

2 2 14 18 17 24 19

Fast SAR De (Gy) 46 93 294 379 402 467 522

      

2 5 20 11 15 32 25

SC-MAR De (Gy) 54  105  299  389  440  415  374 

2 3 19 56 73 68 39

MAR protocols (Table 2). With the exception of the oldest sample (PT5) De values measured using the different laboratory protocols agree with each other within 2s error (Fig. 8), as would be expected given the similarities in laboratory constructed DRCs. The error associated with SC-MAR De values is highly dependent on the variance in Ln/Tn values for a sample, while the single aliquot measurements presented here have lower errors, highlighting the advantage of incorporating inter-aliquot variability in De determination by constructing individual DRCs. When the measured De values for the different protocols are plotted against the expected De values (Fig. 8), the measured and expected sets of data rapidly diverge above 100 Gy. The data in Fig. 8 can be well fitted with a single saturating exponential function (R2 ¼ 0.99 for net integrated SAR measurements). This relationship between the measured and expected De values can be used to identify the maximum limit of reliable De determination for the Luochuan loess section. The saturated exponential function fitted in Fig. 8 deviates from the 1:1 line by 10% around 150 Gy. Based on these observations, we suggest caution in dating samples with equivalent doses greater than 150 Gy that have laboratory constructed DRCs described by either double saturating exponential or saturating exponential plus linear functions as beyond this limit De values based on interpolation onto laboratory generated DRCs are likely to underestimate the true age. This recommended upper limit agrees with previous studies that have found age underestimation at high doses at this site (Buylaert et al., 2007; Lai, 2009).

1051

6. Discussion and conclusions Seven samples from a sedimentary profile with similar source material and a framework of independent age control have been used to construct a natural DRC and investigate whether laboratory constructed DRCs accurately reproduce natural OSL signal growth across a range of doses. Two differences are observed between the natural- and laboratory-constructed DRCs: firstly, the shapes of the DRCs are different (Fig. 3), and secondly, Lx/Tx values continue to grow at high laboratory doses (e.g. above 500 Gy) where it has ceased growing in equivalent natural Ln/Tn doses (e.g. Fig. 3). These differences are still apparent after fitting LM-OSL signals to isolate the fast component, and constructing SC-MAR DRCs. This suggests that the observed differences are not the result of different components or inter-regenerative dose cycle sensitivity changes. From our experiments it is not possible to reject the hypothesis that differences between natural- and laboratory-constructed DRCs may be due to sensitivity change occurring during measurement of the natural signal as suggested by Singhvi et al. (2011). On the basis of the results obtained in this study, the current maximum limit of OSL dating at the Luochuan section using the SAR protocol on quartz grains is 150 Gy, as De estimates above this value are likely to be underestimations (Fig. 8). Further research aimed at investigating the shape of dose response curves or testing protocols for extending the age range may benefit from comparing natural and laboratory constructed dose response curves, in order to validate the assumption that one can accurately reproduce in the laboratory the same luminescence processes that occur in nature.

Acknowledgements This material is based on work supported by the National Science Foundation Graduate Research Fellowship under Grant No. 1053735 to MSC. Aberystwyth University and the Institute of Geography and Earth Sciences are thanked for financially supporting MSC at the International Luminescence and ESR conference in Torun in 2011 where this work was first presented. HMR acknowledges a Leverhulme grant awarded to B.A. Maher which made it possible to collect some of the samples. Aberystwyth Luminescence Research Laboratory benefits from the Climate Change Consortium of Wales (C3W). We also acknowledge the One-Hundred Talent Project of CAS granted to ZPL (A0961).

References

Fig. 8. Measured equivalent dose (De) values as a function of the De values expected for each sample. The solid 1:1 line is flanked by dashed grey lines representing 10% error. The measured De values can be fitted with a single saturating exponential function with an R2 value of 0.99.

Banerjee, D., Murray, A.S., Bøtter-Jensen, L., Lang, A., 2001. Equivalent dose estimation using a single aliquot of polymineral fine grains. Radiat. Meas. 33, 73e94. Bateman, M.D., Frederick, C.D., Jaiswal, M.K., Singhvi, A.K., 2003. Investigations into the potential effects of pedoturbation on luminescence dating. Quaternary Sci. Rev. 22, 1169e1176. Bøtter-Jensen, L., Andersen, C.E., Duller, G.A.T., Murray, A.S., 2003. Developments in radiation, stimulation and observation facilities in luminescence measurements. Radiat. Meas. 37, 535e541. Buylaert, J.P., Vandenberghe, D., Murray, A.S., Huot, S., De Corte, F., Van den Haute, P., 2007. Luminescence dating of old (>70 ka) Chinese loess: a comparison of single-aliquot OSL and IRSL techniques. Quat. Geochronol. 2, 9e14. Choi, J.H., Duller, G.A.T., Wintle, A.G., 2006. Analysis of quartz LM-OSL curves. Ancient TL 24, 9e20. Ding, Z.L., Derbyshire, E., Yang, S.L., Yu, Z.W., Xiong, S.F., Liu, T.S., 2002. Stacked 2.6Ma grain size record from the Chinese loess based on five sections and correlation with the deep-sea d18O record. Paleoceanography 17, 1033. Duller, G.A.T., 2003. Distinguishing quartz and feldspar in single grain luminescence measurements. Radiat. Meas. 37, 161e165. Forman, S.L., 1991. Late Pleistocene chronology of loess deposition near Luochuan, China. Quatern. Res. 36, 19e28. Jain, M., Murray, A.S., Bøtter-Jensen, L., 2003. Characterisation of blue-light stimulated luminescence components in different quartz samples: implications for dose measurement. Radiat. Meas. 37, 441e449.

1052

M.S. Chapot et al. / Radiation Measurements 47 (2012) 1045e1052

Kim, J.C., Duller, G.A.T., Roberts, H.M., Wintle, A.G., Lee, Y.I., Yi, S.B., 2009. Dose dependence of thermally transferred optically stimulated luminescence signals in quartz. Radiat. Meas. 44, 132e143. Kukla, G., An, Z., 1989. Loess stratigraphy in central China. Palaeogeogr. Palaeoclimatol. Palaeoecol. 72, 203e225. Lai, Z.-P., Brückner, H., Zöller, L., Fülling, A., 2007. Existence of a common growth curve for silt-sized quartz OSL of loess from different continents. Radiat. Meas. 42, 1432e1440. Lai, Z., 2006. Testing the use of an OSL standardised growth curve (SGC) for De determination on quartz from the Chinese Loess Plateau. Radiat. Meas. 41, 9e16. Lai, Z., 2009. Chronology and the upper dating limit for loess samples from Luochuan section in the Chinese Loess Plateau using quartz OSL SAR protocol. J. Asian Earth Sci. 37, 176e185. Lai, Z., Murray, A.S., Bailey, R.M., Huot, S., Bøtter-Jensen, L., 2006. Quartz red TL SAR equivalent dose overestimation for Chinese loess. Radiat. Meas. 41, 114e119. Lai, Z.P., Bruckner, H., 2008. Effects of feldspar contamination on equivalent dose and the shape of growth curve for OSL of silt sized quartz extracted from Chinese loess. Geochronometria 30, 49e53. Lai, Z.P., Zöller, L., Fuchs, M., Brückner, H., 2008. Alpha efficiency determination for OSL of quartz extracted from Chinese loess. Radiat. Meas. 43, 767e770. Liu, T., 1985. Loess and the Environment. China Ocean Press, Beijing, China. Liu, T., Chang, T.-h., 1964. The ‘Huangtu’ (Loess) of China. 6th INQUA Congress, Warsaw, pp. 503e524. Lowick, S.E., Preusser, F., 2011. Investigating age underestimation in the high dose region of optically stimulated luminescence using fine grain quartz. Quat. Geochronol. 6, 33e41. Lowick, S.E., Preusser, F., Wintle, A.G., 2010. Investigating quartz optically stimulated luminescence dose-response curves at high doses. Radiat. Meas. 45, 975e984. Lu, Y.C., Wang, X.L., Wintle, A.G., 2007. A new OSL chronology for dust accumulation in the last 130,000 yr for the Chinese Loess Plateau. Quat. Res. 67, 152e160. Murray, A., Buylaert, J.-P., Henriksen, M., Svendsen, J.-I., Mangerud, J., 2008. Testing the reliability of quartz OSL ages beyond the Eemian. Radiat. Meas. 43, 776e780.

Murray, A.S., Svendsen, J.I., Mangerud, J., Astakhov, V.I., 2007. Testing the accuracy of quartz OSL dating using a known-age Eemian site on the river Sula, northern Russia. Quat. Geochronol. 2, 102e109. Murray, A.S., Wintle, A.G., 2003. The single aliquot regenerative dose protocol: potential for improvements in reliability. Radiat. Meas. 37, 377e381. Pawley, S.M., Toms, P., Armitage, S.J., Rose, J., 2010. Quartz luminescence dating of Anglian Stage (MIS 12) fluvial sediments: comparison of SAR age estimates to the terrace chronology of the Middle Thames valley, UK. Quat. Geochronol. 5, 569e582. Roberts, H.M., 2007. Assessing the effectiveness of the double-SAR protocol in isolating a luminescence signal dominated by quartz. Radiat. Meas. 42, 1627e 1636. Roberts, H.M., Duller, G.A.T., 2004. Standardised growth curves for optical dating of sediment using multiple-grain aliquots. Radiat. Meas. 38, 241e252. Roberts, H.M., Wintle, A.G., 2001. Equivalent dose determinations for polymineralic fine-grains using the SAR protocol: application to a Holocene sequence of the Chinese Loess Plateau. Quat. Sci. Rev. 20, 859e863. Singarayer, J.S., Bailey, R.M., 2003. Further investigations of the quartz optically stimulated luminescence components using linear modulation. Radiat. Meas. 37, 451e458. Singhvi, A., Stokes, S., Chauhan, N., Nagar, Y., Jaiswal, M., 2011. Changes in natural OSL sensitivity during single aliquot regeneration procedure and their implications for equivalent dose determination. Geochronometria 38, 231e241. Stevens, T., Armitage, S.J., Lu, H., Thomas, D.S.G., 2007. Examining the potential of high sampling resolution OSL dating of Chinese loess. Quat. Geochronol. 2, 15e22. Timar-Gabor, A., Vasiliniuc, Å., Vandenberghe, D.A.G., Cosma, C., Wintle, A.G. Investigations into the reliability of SAR-OSL equivalent doses obtained for quartz samples displaying dose response curves with more than one component. Radiat. Meas., in press. Wang, X.L., Lu, Y.C., Wintle, A.G., 2006. Recuperated OSL dating of fine-grained quartz in Chinese loess. Quat. Geochronol. 1, 89e100. Wintle, A.G., Murray, A.S., 2006. A review of quartz optically stimulated luminescence characteristics and their relevance in single-aliquot regeneration dating protocols. Radiat. Meas. 41, 369e391.