testing optically stimulated luminescence dating of sand ...tor/documents/epsl2001.pdf19098,...

Testing optically stimulated luminescence dating ofsand-sized quartz and feldspar from £uvial deposits

J. Wallinga a;b;c;*, A.S. Murray b, G.A.T. Duller c, T.E. To«rnqvist d

a The Netherlands Centre for Geo-ecological Research (ICG), Faculty of Geographical Sciences, Utrecht University, P.O. Box 80115,NL-3508 TC Utrecht, The Netherlands

b Nordic Laboratory for Luminescence Dating, Department of Earth Sciences, Aarhus University, RisÖ National Laboratory,DK-4000 Roskilde, Denmark

c Institute of Geography and Earth Sciences, University of Wales, Aberystwyth SY23 3DB, UKd Department of Earth and Environmental Sciences, University of Illinois at Chicago, 845 West Taylor Street, Chicago,

IL 60607-7059, USA

Received 17 May 2001; received in revised form 22 August 2001; accepted 30 September 2001

Abstract

We apply single-aliquot optically stimulated luminescence (OSL) dating to quartz- and feldspar-rich extracts fromfluvial channel deposits of the Rhine^Meuse system in The Netherlands. The time of deposition of these deposits istightly constrained by radiocarbon dating or historical sources. This allows us to compare OSL ages obtained on quartzand infrared OSL (IR-OSL) ages obtained on potassium-rich feldspar with independent ages over the range of 0.3^13ka. We show that the quartz OSL ages are in good agreement with the expected age. Using IR-OSL dating of feldspar,we find a slight age overestimate for the youngest sample, whereas for older samples the age is significantlyunderestimated. We also apply OSL dating to older fluvial and estuarine channel deposits with limited independentchronological constraints. Comparison of feldspar IR-OSL ages with the quartz OSL ages up to V200 ka shows a cleartrend, where the former severely underestimates the latter. This trend is similar to that found for the samples withindependent age control, indicating that the feldspar IR-OSL ages are erroneously young for the entire age range. In theyoungest samples, incomplete resetting of the IR-OSL signal prior to deposition probably masks the ageunderestimation. We show that the IR-OSL age underestimation is partly caused by changes in trapping probabilitydue to preheating. Correction for this phenomenon improves the IR-OSL ages slightly, but does not provide a completesolution to the discrepancy. We suggest that, in the light of the problems encountered in the IR-OSL dating of feldspar,quartz is the mineral of choice for OSL dating of these deposits. However, feldspar dating should continue to beinvestigated, because it has potential application to longer time scales. ß 2001 Elsevier Science B.V. All rightsreserved.

Keywords: optically stimulated luminescence dating; quartz; feldspar group; stream sediments; age

1. Introduction

Optically stimulated luminescence (OSL) datingis a rapidly developing technique that providesabsolute chronologies for late Quaternary clastic

0012-821X / 01 / $ ^ see front matter ß 2001 Elsevier Science B.V. All rights reserved.PII: S 0 0 1 2 - 8 2 1 X ( 0 1 ) 0 0 5 2 6 - X

* Corresponding author. Tel. : +31-30-253-2749;Fax: +31-30-253-1145.

E-mail address: [email protected] (J. Wallinga).

EPSL 6028 4-12-01

Earth and Planetary Science Letters 193 (2001) 617^630

www.elsevier.com/locate/epsl

sedimentary records [1]. OSL dating of the sandyfraction can be applied to quartz up to an age of100^200 ka, or to feldspar, which o¡ers the po-tential to extend the age range to 1 Ma. Fluvialrecords now represent one of the most commonsedimentary environments dated by OSL meth-ods. However, a potential problem in £uvial sys-tems is that light exposure of sand during trans-port may not have been su¤cient to completelyreset the OSL signal of all grains prior to deposi-tion. Incomplete resetting of the OSL signal willlead to an overestimation of the age. Most studiesapplying OSL dating to £uvial deposits have fo-cused on chronologically unconstrained, or poorlyconstrained, Pleistocene successions. With the ex-ception of a few studies (notably [2^5]), little ef-fort has been made to provide a sound justi¢ca-tion for the use of luminescence dating in £uvialsettings.

Rigorous comparison of OSL ages with inde-pendent ages is only possible if the latter arehighly accurate and if the stratigraphic relation-ship between OSL samples and independent agesis indisputable. Studies must also provide enoughbackground information to assess the quality ofluminescence and independent ages. For coarse-grain quartz OSL dating, several comparativestudies that meet these guidelines have been con-ducted in aeolian environments [6,7]. For coarse-grain feldspar infrared OSL (IR-OSL) dating, onthe other hand, validation of the technique is sur-prisingly limited; few studies ¢t the criteria forrigorous comparison outlined above. Good agree-ment with independent age control was found forsamples younger than 30 ka for aeolian dunesands from New Zealand [8]. On the otherhand, IR-OSL ages on samples from the sametype of deposit from Germany showed an under-estimation of age [9].

In the present study, we apply OSL dating toboth quartz and feldspar extracts from sandychannel deposits from the Rhine^Meuse systemin The Netherlands. We sampled four channelbelts for which the period of activity is accuratelyknown from historical sources or radiocarbondating. Comparison of OSL dating results fromthese samples with unusually tight independentage constraints presents an unparalleled opportu-

nity to test the accuracy of OSL dating resultsfrom submodern (0.3 ka) to Late Weichselian(V13 ka) £uvial sediments. Additionally, OSLdating of samples from chronologically uncon-strained £uvial and estuarine channel depositsfrom the same area allows comparison of quartzand feldspar OSL ages up to an age of V200 ka.

2. Study area and independent age constraints

The Rhine^Meuse Delta (Figs. 1 and 2) is lo-cated in the southeastern part of the North SeaBasin. The Holocene delta is underlain by Weich-selian (oxygen-isotope stages (OIS) 2^4), sandy togravelly £uvial channel deposits. The HoloceneRhine^Meuse Delta was formed in response torelative sea-level rise [10,11] and contains wide-

Fig. 1. The Rhine^Meuse Delta in The Netherlands and lo-cation of sampling sites.

EPSL 6028 4-12-01

J. Wallinga et al. / Earth and Planetary Science Letters 193 (2001) 617^630618

spread in situ organics interbedded with £uvialclastics, thus enabling highly accurate 14C chro-nologies to be established. We collected OSL sam-ples from three Holocene channel belts (Waal,Linge, and Schaik systems) and one Late Weich-

selian channel belt (Elden core), and additionallyfrom Weichselian to Saalian (OIS 2 and older),£uvial and estuarine channel deposits (Leidschen-dam core) (Figs. 1 and 2).

According to historical maps, the deposits atthe sampling site of the Waal (Winssen core)were formed between AD 1688 and 1723 [12],which corresponds to an age of V300 yr (notethat we use AD 2000 as a reference for calendarages). The Winssen sample was collected to assessthe potential e¡ects of poor bleaching. We se-lected a 300-yr-old sample rather than a modernone to avoid the in£uence of locks and water-works that have altered sediment transport inthe present-day river, and thus the bleaching en-vironment.

The ages of the other two Holocene channelbelts are well constrained by 14C dating (Table1) with periods of activity no longer than 1700yr (Linge system) and 800 yr (Schaik system).Radiocarbon dating [13] employed acceleratormass spectrometry (AMS) of terrestrial botanicalmacrofossils [14]. For both the Linge and theSchaik systems, a larger number of 14C ages areavailable; we used those that are located closestto the sites of our OSL samples. The beginning ofactivity of both channel belts was determined by14C dating the top of peat directly underlying

Fig. 2. Schematic east^west cross section showing relative po-sitioning of the samples. The Winssen, Rumpt and Schellui-nen samples were taken from sandy £uvial channel depositswithin the Holocene Rhine^Meuse Delta. The Elden andLeidschendam samples were taken from Pleistocene, predom-inantly £uvial channel deposits underlying the Holocene del-ta. Sample location (local coordinates) and depth (meters be-low the surface): Winssen (168.485/435.475, 32.75 m),Rumpt I-3 (139.355/433.880, 34.35 m), Rumpt IV-2(139.920/433.330, 32.95 m), Schelluinen II-3 (123.290/429.875, 33.15 m), Schelluinen II-6 (123.290/429.875, 35.55m), Elden (187.985/440.000, 36.25 m), Leidschendam(87.540/454.380, for sample depths see [20]).

Table 1Independent age constraints of OSL-dated £uvial channel deposits

Fluvial system OSL samples 14C age Calendar ageb

(yr BP) (yr before AD 2000)(median and 2c con¢dence interval)

Beginning End Beginning End

Waal Winssen 312c 277c

Linge Rumpt I-3, Rumpt IV-2 2 235 þ 35a1 2 300d1 (2 190^2 370) V700e

Schaik Schelluinen II-3, Schelluinen II-6 5 050 þ 85a2 4 605 þ 45a3 5 850d1 (5 670^6 030) 5 360d2 (5 210^5 470)Late Weichselian Elden 11 063 þ 12a4 13 155f (12 990^13 320)a Laboratory numbers: (1) UtC-1481/1482; (2) UtC-1144/1300; (3)UtC-1128/1129/1130/1141/1142 (all based on AMS 14C mea-surements of terrestrial plant macrofossils, except for samples UtC-1128, 1141 and 1481 [13]); (4) Hd-19607, Hd-18648, Hd-19098, Hd-19092, Hd-18622, Hd-19037, Hd-18438 (AMS 14C measurements of decadal samples of poplar buried by the LaacherSee Tephra [16]).b As derived from historical sources or calibrated 14C ages. For the latter 50 yr was added to each age (initially determined in yrBP = AD 1950) to enable direct comparison with OSL ages.c Obtained from historical maps [12].d Calibration of 14C ages according to the Groningen CAL25 program [18] using smoothed curves [19]; smoothing parametersused: (1) 100; (2) 200.e Historical age of damming of the river at its upstream bifurcation [15].f Following calibration by Friedrich et al. [16], including a systematic uncertainty of 70 yr.

EPSL 6028 4-12-01

J. Wallinga et al. / Earth and Planetary Science Letters 193 (2001) 617^630 619

overbank deposits (sampling strategy is exten-sively discussed by To«rnqvist and Van Dijk[13]). The end of activity of the Linge systemwas a result of arti¢cial damming at its upstreambifurcation around AD 1300 [15]. The end of ac-tivity of the Schaik system was derived from theage of the base of peat overlying overbank depos-its and basal peat in the residual channel [13].

We also sampled channel deposits of LateWeichselian age, which underlie the Holocene del-taic deposits at Elden (Figs. 1 and 2). The £uvialsands at this location are extremely rich in pumicethat originated from the Laacher See volcaniceruption in the Eifel (Germany), 14C dated at11 063 þ 12 yr BP [16]. Pumice from the LaacherSee eruption blocked the River Rhine for severaldays [17], and was transported downstream whenthe dam collapsed. The pumice has been foundthroughout Late Weichselian £uvial deposits inThe Netherlands [15], and the high concentration(up to 25% by volume) at our sampling site makesit highly likely that these sands were depositedshortly after the volcanic eruption.

In addition to the sites with independent agecontrol (summarized in Table 1), samples weretaken from a core retrieved from Saalian (OIS6^8) to Weichselian (OIS 2^4) £uvial and estua-rine channel deposits near Leidschendam [20](Figs. 1 and 2). OSL dating of the quartz andfeldspar separates from this core allows us tocompare the results obtained on both mineralfractions up to an age of V200 ka.

3. Methods

Most samples were taken using a simple hand-operated suction corer, enabling us to obtain 30-cm-long samples in opaque PVC tubes [21]. TheLeidschendam core was obtained using a mecha-nized bailer drilling unit [22] of the NetherlandsInstitute of Applied Geoscience TNO, yielding anundisturbed core with 10 cm diameter. The coreswere opened in subdued red light, after whichsamples for OSL dating were taken. All sampleswere water-washed and treated with 10% HCl and30% H2O2 to remove carbonates and organic ma-terial. After drying, the samples were sieved and

subsequently density-separated using an aqueoussolution of sodium polytungstate to extract thepotassium-rich feldspar fraction lighter than 2.58g/cm3. The denser fraction was treated with con-centrated hydro£uoric acid for 40 min to obtain aclean quartz sample and to etch away the outer 10Wm of the quartz grains. No hydro£uoric acidetching was used for the feldspar.

Measurements were made on an automatedRisÖ TL/OSL reader, using an internal 90Sr/90YL-source [23]. The sample grains were mountedon aluminum or stainless steel discs using siliconespray. Blue light emitting diodes were used forstimulation of quartz (at 125³C) and the resultingluminescence signal was detected through 9 mmof Schott U-340 ¢lters (detection window 250^390nm). The single-aliquot regenerative dose (SAR)protocol [24] was used for estimation of the equiv-alent dose. A relatively low preheat (200³C for 10s) was used to avoid thermal transfer e¡ects [1,25]that were shown to a¡ect the equivalent dose ofthe Winssen sample when more stringent preheatswere used (Fig. 3a). For the Leidschendam sam-ples a more usual 10 s preheat at 260³C was usedbecause here thermal e¡ects are expected to benegligible. This was con¢rmed by the preheat pla-teau obtained for sample Leidschendam I (Fig.3b). The test-dose response was measured afterheating to 160³C for all samples.

For the feldspar separates, we used Schott BG-39 and Corning 7-59 ¢lters, giving a transmissionwindow between 320 and 480 nm. The single-ali-quot additive dose (SAAD) procedure [26,27],with a 10 min preheat at 220³C, was used forestimation of the equivalent dose. Optical stimu-lation was provided by infrared diodes (emittinground 880 nm). The equivalent dose was also es-timated using the SAR protocol for feldspar [28],using a 10 s preheat at 290³C for natural andregenerative doses, and heating to 210³C for thetest doses. An infrared laser diode (emitting at830 nm) was used for stimulation.

The natural dose rate was estimated in the lab-oratory using high-resolution Q-spectrometry [29]on bulk samples (results in Table 2), that weretaken from around the sample used for equiva-lent-dose determination. It is fair to assume thatthe samples have been saturated with water

EPSL 6028 4-12-01


throughout their lifetimes, which diminishes thedose rate [38].

4. OSL dating results and discussion

4.1. Quartz OSL

The quartz OSL dating results and independentages are presented in Table 2 and Fig. 4. For theyoungest sample (Winssen), the OSL age slightlyoverestimates the known historical age of thesample. For the other samples with independentage control all OSL ages are in excellent agree-

ment with the radiocarbon-dated periods of activ-ity. The slight o¡set (600 yr) found for the Wins-sen sample is likely caused by incompletebleaching, and is comparable to that found formodern channel deposits in other parts of theworld [2,4,5]. We observed that application ofhigher preheat temperatures (e.g. 260³C) resultedin a greater o¡set for the Winssen sample (Fig.3a), and also in a slight overestimation of age forthe other samples with independent age control(not shown). These results are in accordancewith those of Rhodes [25] and indicate that strin-gent preheats should be avoided when datingyoung samples that might not have been thor-oughly exposed to light prior to deposition.

In addition to our test of the validity of quartzOSL dating, we applied OSL dating to older de-posits with limited age constraints to allow com-parison of quartz OSL and feldspar IR-OSL dat-ing results for a wider age range. The geologicalcontext and the quartz OSL dating results onthese samples (Leidschendam core) are discussedby To«rnqvist et al. [20] ; a brief summary is givenhere. The OSL data suggest that the base of thesuccession (samples VIII^X; Table 2) was depos-ited during the Saalian glaciation (OIS 6). The£attening of the OSL dose^response curve at therelatively high doses these samples have absorbedampli¢es the scatter in the OSL measurements,and this is re£ected in the relatively large uncer-tainties in the age estimates. The OSL age of sam-ple VII suggests an Eemian or Early Weichselian(OIS 5) origin, which is supported by the ¢rstoccurrence of transgressive marine shell remainsat this level [20].

OSL ages for the upper part of the succession(samples I^VI) point to deposition around OIS 4,i.e. during a period with relatively low sea level.However, mud drapes containing warm pollenand estuarine diatoms were encountered near thebase of this unit (sample VI, see also [20]), sug-gesting that these sediments must have been de-posited during a period of relatively high sea level.Hence, we cannot preclude the possibility thatthese deposits were formed during the precedingsea-level highstand, i.e. OIS 5a (V80 ka), andthat they are possibly older than the quartz OSLages suggest. The slight age reversal for samples

Fig. 3. Equivalent doses obtained for a range of preheat tem-peratures (each held for 10 s) for aliquots of quartz separatefrom the Winssen (a) and Leidschendam I (b) samples. Eachestimate is the average from three aliquots, and error barsindicate the standard error on the mean. The equivalent dose(circles), and the recycling ratio (triangles) [24] are bothshown. The equivalent-dose estimate used for age determina-tion (Table 2; Winssen preheat 200³C, Leidschendam I pre-heat 260³C) is indicated by the dotted line.

EPSL 6028 4-12-01


Leidschendam III^VI might also indicate that theOSL ages obtained on samples V and VI (V60ka) could be too young. It is interesting to notethat slight age underestimation was also reportedfor quartz OSL ages of Eemian (OIS 5e) depositsfrom Denmark [39].

This study for the ¢rst time rigorously com-pares OSL ages and independent age control for£uvial deposits over a relatively wide age range(up to 13 ka), and the results underline the val-idity of quartz OSL dating for establishing abso-lute chronologies for £uvial deposits. However,the controversy over the age of the Weichseliandeposits in the Leidschendam core stresses theneed for comparisons of OSL ages with independ-ent age control for pre-Holocene deposits. Futureresearch should focus on such comparisons to fur-ther increase the con¢dence in quartz OSL ages.

4.2. Feldspar IR-OSL

The feldspar IR-OSL dating results obtained bythe SAAD protocol [26,27] are presented in Table3 and Fig. 4. For the Winssen sample the IR-OSLage overestimates the independent age and is alsoslightly higher than that obtained by OSL datingof the quartz separate. For the Rumpt samplesIR-OSL ages are found in agreement with theindependent age range and with the quartz OSLdating results. However, for the Schelluinen andElden samples an IR-OSL age underestimation isfound, both compared to independent ages andcompared to the quartz OSL ages. An increasingdeviation with increasing age (Fig. 4b) suggeststhat the underestimation is a relative e¡ect, pos-sibly masked in the younger samples by poorbleaching. This underestimation of IR-OSL age

Table 2Quartz OSL dating results

Sample Grainsize

Radionuclide concentrationa Doseratea;b

Equivalentdose

OSLage

Independentage

(Wm) (Bq/kg) (Gy/ka) (Gy) (ka) (ka)238U 226Ra 232Th 40K

Winssen 180^212 12 þ 2 10.8 þ 0.2 10.9 þ 0.2 362 þ 7 1.27 þ 0.05 1.17 þ 0.12c 0.92 þ 0.10 0.3Rumpt I-3 180^212 15 þ 2 10.1 þ 0.4 12.3 þ 0.4 404 þ 13 1.36 þ 0.05d 1.67 þ 0.13d 1.23 þ 0.10d 0.7^2.4Rumpt IV-2 180^212 35 þ 5 31.7 þ 0.5 33.0 þ 0.5 482 þ 10 2.10 þ 0.14 3.7 þ 0.2 1.75 þ 0.10 0.7^2.4Schelluinen II-3 180^212 12 þ 3 9.9 þ 0.3 10.7 þ 0.3 398 þ 8 1.33 þ 0.05 6.9 þ 0.5 5.1 þ 0.4 5.2^6.0Schelluinen II-6 180^212 11 þ 2 9.3 þ 0.2 9.9 þ 0.2 363 þ 7 1.22 þ 0.09d 7.5 þ 0.2d 6.1 þ 0.5d 5.2^6.0Elden 180^212 13 þ 2 14.3 þ 0.4 16.4 þ 0.4 415 þ 12 1.51 þ 0.05 20.0 þ 1.0 13.3 þ 0.8 13.0^13.3Leidschendam I 180^212 10 þ 3 9.5 þ 0.5 9.6 þ 0.4 297 þ 13 1.11 þ 0.06 54 þ 3 48 þ 4 ^Leidschendam II 180^212 6 þ 3 9.5 þ 0.3 10.8 þ 0.3 249 þ 7 0.99 þ 0.05 54 þ 5 55 þ 6 ^Leidschendam III 180^212 6 þ 2 9.5 þ 0.4 7.5 þ 0.3 200 þ 10 0.78 þ 0.05 64 þ 6 82 þ 9 ^Leidschendam IV 180^212 12 þ 3 12.8 þ 0.3 14.1 þ 0.3 341 þ 7 1.29 þ 0.06 92 þ 7 71 þ 6 ^Leidschendam V 180^212 6 þ 3 6.9 þ 0.5 8.2 þ 0.4 253 þ 13 0.92 þ 0.05 56 þ 3 61 þ 5 ^Leidschendam VI 180^212 8 þ 3 9.8 þ 0.2 10.7 þ 0.2 314 þ 7 1.13 þ 0.05 66 þ 3 58 þ 4 ^Leidschendam VII 180^250 12 þ 3 6.8 þ 0.3 6.7 þ 0.2 237 þ 6 0.86 þ 0.04 104 þ 6 120 þ 9 ^Leidschendam VIII 180^250 10 þ 2 6.5 þ 0.4 8.2 þ 0.4 201 þ 10 0.79 þ 0.05 124 þ 7 158 þ 13 ^Leidschendam IX 180^212 11 þ 2 5.5 þ 0.3 7.4 þ 0.2 190 þ 7 0.74 þ 0.04 107 þ 10 145 þ 16 ^Leidschendam X 180^212 7 þ 3 6.9 þ 0.3 8.2 þ 0.2 240 þ 6 0.87 þ 0.04 156 þ 23 180 þ 28 ^a Spectral data from high-resolution Q-spectroscopy converted to activity concentrations and in¢nite matrix dose rates using theconversion data given by Olley et al. [30].b The natural dose rate was calculated from the in¢nite matrix dose rate using attenuation factors given by Mejdahl [31], and in-cludes a contribution from cosmic rays [32]. A contribution from internal K dose was calculated based on U and Th contents re-ported by Mejdahl [33], using an a-value of 0.04 þ 0.01 [34], which resulted in an internal dose rate of 0.028 þ 0.013 Gy/ka. Alldose rates calculated for a water content of 20 þ 2% (based on a porosity of 34 þ 3% [35]) using attenuation factors given by Zim-merman [36].c Four outliers (De s 6 Gy) from a total of 34 aliquots were not incorporated.d Values di¡er slightly from those reported by Wallinga and Duller [37] due to a small shift in the source calibration, and an im-proved water content estimation.

EPSL 6028 4-12-01


is unlikely to be caused by errors in determinationof the external dose rate; this would also a¡ectthe quartz ages. It is improbable that the di¡er-ences between OSL ages on quartz and IR-OSLages on feldspar are caused by di¡erences inbleaching, as the quartz signal is more readilyreset by sunlight [1], and because the quartzages agree with the independent chronology.

To investigate whether the age underestimationcould be caused by sensitivity changes during themeasurement procedure, a SAR protocol for feld-spar [28] was employed. Using this protocol, agessimilar to those obtained using the SAAD proce-dure were obtained (Table 3, see also [28]). Anadditional advantage of the SAR procedure isthat it is straightforward to test whether preheat-ing removes all unstable trapped charge caused bylaboratory irradiation. Using the SAR protocol,we found equivalent doses to be independent ofpreheat temperature for 10 s preheats above200³C (Fig. 5).

The IR-OSL age of feldspar separates from theLeidschendam core was measured using the SARprotocol. The average equivalent dose of three tosix aliquots of each sample is shown in Table 3. InFig. 6 the IR-OSL ages obtained on the samplesare plotted as a function of the quartz OSL age.All but two samples follow a trend where thefeldspar IR-OSL age is only half that of thequartz OSL age from the same sample. For thetwo samples that do not follow this trend (Leid-schendam VII and VIII), an atypically large scat-ter was observed between equivalent doses ob-tained on di¡erent feldspar aliquots, suggestingthat the IR-OSL signal was not completely resetfor all grains at the time of deposition [41].

From the results for the samples with independ-ent age control, we deduced that the accuracy ofthe quartz OSL ages is superior to that of thefeldspar IR-OSL ages. Combining the results ob-tained on the samples with independent age con-trol, and that from the Leidschendam samples, weconclude that the IR-OSL ages obtained on thepotassium-rich feldspar separates severely under-estimate the age of our samples, but that in some

Fig. 4. (a) OSL ages obtained on quartz separates using theSAR protocol [24] with a 10 s 200³C preheat (¢lled circles),and IR-OSL ages obtained on potassium-rich feldspar sepa-rates using the SAAD procedure [26,27] (open circles) plottedagainst the independent age of the samples. All errors indi-cate 2c con¢dence intervals. (b) Di¡erence between the lumi-nescence ages and the independent age control. Errors in-clude uncertainties in both the independent ages and theOSL ages.

Fig. 5. Equivalent doses obtained for a range of preheat tem-peratures (each held for 10 s) for aliquots of feldspar sepa-rate from sample Schelluinen II-3. The sample was heated to150³C after the test dose was administered. Each estimate isthe average of three aliquots. The equivalent dose obtainedby the SAR procedure (¢lled circles) and the recycling ratio(¢lled triangles) are indicated, as are the correction factor fordi¡erent preheat temperatures (open triangles) and the equiv-alent dose after correction with this factor (open circles). Thederivation of the correction procedure is described in themain text.

EPSL 6028 4-12-01


cases the underestimation is masked by incom-plete bleaching. The quartz OSL ages of the sam-ples from the Leidschendam core might slightlyunderestimate the true age, as was discussed inSection 4.1. If this is the case, the feldspar IR-OSL age underestimation is even more severethan indicated by the comparison of quartz OSLand feldspar IR-OSL results. In the following sec-tions, the causes for the feldspar IR-OSL ageunderestimation are considered.

5. Possible reasons for age underestimation infeldspar IR-OSL

5.1. Anomalous fading

Anomalous fading is the loss of electrons from

traps on a time scale that is short compared withthe lifetime predicted on the basis of their trapdepth. This phenomenon is only known to a¡ectfeldspars and gives rise to an age underestimation[42^44]. We carried out fading tests in three di¡er-ent ways.

Firstly, the decay in IR-OSL was monitoredfollowing laboratory irradiation of previously un-measured natural samples. Doses similar to thenatural dose of the sample were used for this ex-periment. After irradiation, the samples were pre-heated to 220³C for 10 min and measurementswere made using short exposure to infrared light.To correct for the decay due to this measurement,the same measurements were made on naturalsamples that did not receive a laboratory dosebut were otherwise treated identically. The fadingratio is given by the ratio of irradiated to natural

Table 3Feldspar IR-OSL dating results

Sample InternalKa

Doserateb

Equivalent dose IR-OSLage

Independentage

(%) (Gy/ka) (Gy) (ka) (ka)

SAADc SARd SAADc SARd

Winssen 8.5 1.88 þ 0.08 3.0 þ 0.2 2.9 þ 0.2 1.58 þ 0.14 1.52 þ 0.14 0.3Rumpt I-3 7.4 1.92 þ 0.08e 2.38 þ 0.05e 2.8 þ 0.4 1.24 þ 0.05e 1.5 þ 0.2 0.7^2.4Rumpt IV-2 9.5 2.87 þ 0.13 4.53 þ 0.13 4.2 þ 0.3 1.58 þ 0.08 1.47 þ 0.13 0.7^2.4Schelluinen II-3 7.9 1.93 þ 0.07 7.7 þ 0.2 7.4 þ 0.5 4.0 þ 0.2 3.8 þ 0.3 5.2^6.0Schelluinen II-6 9.8 1.91 þ 0.11e 7.9 þ 0.2e 7.6 þ 0.4 4.2 þ 0.3e 4.0 þ 0.3 5.2^6.0Elden 8.0 2.12 þ 0.08 19.6 þ 0.5 16.8 þ 1.2 9.2 þ 0.4 7.9 þ 0.6 13.0^13.3Leidschendam I 9.2 1.68 þ 0.09 ^ 46.9 þ 1.8 ^ 28 þ 2 ^Leidschendam II 8.2 1.51 þ 0.09 ^ 47.6 þ 1.5 ^ 32 þ 2 ^Leidschendam III 9.2 1.35 þ 0.08 ^ 59 þ 4 ^ 44 þ 4 ^Leidschendam IV 9.1 1.87 þ 0.09 ^ 71 þ 5 ^ 38 þ 3 ^Leidschendam V 9.2 1.49 þ 0.08 ^ 52 þ 3 ^ 35 þ 3 ^Leidschendam VI 9.0 1.69 þ 0.08 ^ 65 þ 3 ^ 38 þ 3 ^Leidschendam VII 8.9 1.47 þ 0.12 ^ 146 þ 17 ^ 100 þ 15 ^Leidschendam VIII 8.5 1.36 þ 0.12 ^ 191 þ 13 ^ 140 þ 16 ^Leidschendam IX 10.1 1.36 þ 0.08 ^ 96 þ 6 ^ 71 þ 6 ^Leidschendam X 9.1 1.43 þ 0.07 ^ 112 þ 4 ^ 78 þ 5 ^a The potassium content of the potassium-feldspar separates was determined by L-counting in a GM multicounter system [40]. Acalibration uncertainty of 5% was assumed.b Calculations similar to quartz, apart from a contribution from internal potassium [33]. An a-value of 0.08 þ 0.02 [34] was usedto calculate the contribution from internal K-radiation (0.06 þ 0.03 Gy/ka). As the separates were not HF-etched a small contribu-tion from external K-radiation was also included (typically V0.02 Gy/ka).c SAAD protocol [26,27] used for equivalent-dose estimation.d SAR protocol [28] used for equivalent-dose estimation.e Values di¡er slightly from those reported by Wallinga and Duller [37] due to a small shift in the source calibration, and an im-proved water content estimation.

EPSL 6028 4-12-01


IR-OSL response before storage, divided by thesame ratio after storage at ambient temperaturefor 6 months.

For the second and third tests, fading waschecked using the SAR procedure [28]. We com-pared the sensitivity-corrected OSL signal afterirradiation (V50 Gy) and storage with that mea-sured immediately after irradiation (time betweenirradiation and measurement was a maximum of4 h). These fading tests were performed on ali-quots that had previously been used for equiva-lent-dose determination using the feldspar SARprocedure [28]. The samples were preheated to290³C for 10 s prior to measurement of the IR-OSL signal. Samples were stored either for4 months at ambient temperature (second test),or for 10 days at 100³C (third test). Due to alack of material, it was not possible to apply allthree methods to each sample.

Some fading was detected in the Elden sampleusing the third test (Table 4); unfortunately therewas not enough material to check this fading withthe other two methods. The fading in this samplemight be associated with the recent volcanic ori-gin of some of the feldspar, as the material wasdeposited shortly after the Laacher See volcanic

Fig. 6. Feldspar IR-OSL ages as a function of the quartzOSL ages of the same samples. The SAR procedure wasused for equivalent-dose determination in both (see Tables 2and 3 for sample names, equivalent doses and ages). Nearlyall samples follow a trend where the feldspar IR-OSL age isonly half the quartz OSL age (¢lled circles, dash^dot trendline). The two samples that are indicated by triangles showunusually wide scatter in the IR-OSL equivalent-dose deter-minations and were not incorporated in the regression. Alsoshown are the feldspar IR-OSL ages after application of thecorrection factor as determined for each sample (open sym-bols, dotted trend line; Table 5).

Table 4IR-OSL fading tests on the feldspar separates

Sample Fading ratio

Conventional SAR SAR6 months ambient 4 months ambient 10 days 100³C

Winssen ^ ^ 1.00 þ 0.05Rumpt I-3 0.95 þ 0.04 ^ 1.003 þ 0.013Rumpt IV-2 1.01 þ 0.08 ^ ^Schelluinen II-3 0.96 þ 0.02 ^ 1.06 þ 0.02Schelluinen II-6 0.87 þ 0.03 ^ 1.00 þ 0.02Elden ^ ^ 0.907 þ 0.010Leidschendam I ^ 1.003 þ 0.005 0.940 þ 0.007Leidschendam II ^ 0.98 þ 0.02 1.034 þ 0.002Leidschendam III ^ 1.00 þ 0.02 0.91 þ 0.03Leidschendam IV ^ 0.934 þ 0.012 0.970 þ 0.011Leidschendam V ^ 1.00 þ 0.03 0.962 þ 0.011Leidschendam VI ^ 0.957 þ 0.006 0.965 þ 0.007Leidschendam VII ^ 0.972 þ 0.008 0.962 þ 0.014Leidschendam VIII ^ 0.98 þ 0.02 0.97 þ 0.02Leidschendam IX ^ 0.978 þ 0.006 0.944 þ 0.009Leidschendam X ^ 0.996 þ 0.007 0.936 þ 0.008

Group average 0.95 þ 0.03 0.980 þ 0.007 0.970 þ 0.011Overall average 0.970 þ 0.007

EPSL 6028 4-12-01


eruption [16]. Based on the results of the fadingtests (Table 4), we cannot completely rule out theexistence of anomalous fading in our samples(average fading ratio 0.970 þ 0.007). In interpret-ing this fading ratio, one should also keep in mindthat the laboratory time scale for fading tests isvery short compared to the geological time scale.However, tests on samples that show an IR-OSLage underestimate as a result of anomalous fadingtend to give an unambiguous indication of thepresence of fading [45]. On balance we thereforethink it is unlikely that anomalous fading is themain cause for the severe age underestimationfound for our feldspar samples.

5.2. Sensitivity change

The SAR procedure was developed to over-come problems with sensitivity change duringmeasurement [24,28,46]. However, Murray andWintle [24] pointed out that if sensitivity changesoccur between measurement of the natural OSLsignal and measurement of the OSL from the testdose related to the natural, the SAR procedurewill not detect or correct these. Wallinga et al.[47] have shown that heating of feldspar grainsto temperatures higher than 200³C for 10 s can

cause changes in the charge trapping probability,and thus can change the overall sensitivity. Inboth SAR and SAAD measurement procedures,such a change in trapping probability would oc-cur during preheating of the aliquot (prior to the¢rst measurement of the IR-OSL). Using samplesfrom the same area as discussed here, Wallinga etal. [47] showed that the change in trapping prob-ability resulted in underestimation of a knownlaboratory dose administered prior to any heatingof the sample.

Changes in trapping probability because ofheating can be avoided if laboratory doses aregiven prior to heating of the sample, as is thecase in multiple-aliquot methods. To test forthis, we applied the single-aliquot regenerationand added dose (SARA) procedure [48], whichis (despite its name) a multiple-aliquot procedure.We used the SAR protocol for determination ofthe dose in those aliquots that just retained theirnatural signal and in those where a laboratorydose had been added to the natural dose, therebyslightly modifying the standard SARA procedure[48]. Using this protocol, higher equivalent dosesare obtained than with direct SAR measurements(Fig. 7, Table 5). Although clearly an improve-ment for feldspar separates, the SARA proceduremight not be the preferred protocol. Firstly, it isextremely time consuming, as the method needsequivalent-dose determinations to be carried outon a large number of aliquots. Secondly, theequivalent dose is obtained by extrapolation,which is not desirable, especially for older sam-ples. Finally, a linear extrapolation might not bejusti¢ed [3].

As an alternative to the use of a SARA proce-dure to circumvent the problem, a sample-depen-dent correction factor can be determined by mea-suring the extent of the change in trappingprobability. For this purpose, three aliquotsfrom each sample were bleached for at least 2 hin a Ho«nle SOL2 solar simulator, and subse-quently given a dose similar to their naturaldose. As this ¢rst dose is administered prior toany heating, it is expected to have the same trap-ping sensitivity as the natural dose. Hence, a cor-rection factor for the natural equivalent dose canbe derived by dividing the known laboratory dose

Fig. 7. Equivalent-dose determination on feldspar separatesfrom sample Schelluinen II-6 using the SARA protocol [48].The equivalent dose obtained by this procedure is not af-fected by changes in the trapping probability due to preheat-ing, as the added doses are administered prior to any heatingof the aliquots. In the absence of trapping sensitivitychanges, the regression should follow the dotted line of unitslope.

EPSL 6028 4-12-01


by the dose estimated by the SAR procedure. Us-ing SAR and these correction factors (Table 5),similar ages were obtained as by the SARA pro-tocol (Table 5). An apparently more elegant ap-proach would be to preheat only to temperaturesbelow 200³C for 10 s, but this is outside the tem-perature range needed to remove unstable trappedcharge after laboratory irradiation (indicated bythe rising plateau in this region in Fig. 5).

Either the use of the correction factors or theapplication of the SARA procedure improves thefeldspar IR-OSL ages, and brings them closer tothe independent ages and the quartz OSL ages(Table 5). However, even after correction fortrapping probability there is still a clear ageunderestimation (Fig. 6). It must be recognizedthat the correction factor may be underestimatedas a consequence of accidental heating of the ex-tracts prior to measurement. During sample prep-aration the extracts are exposed to temperaturesabove ambient during drying at 60³C, duringtreatment with H2O2 (some samples were heatedsigni¢cantly by exothermic reactions), and during

bleaching in the solar simulator. It is possible thateven at these relatively low temperatures the trap-ping probability changes. In radioluminescencestudies of potassium feldspar, sensitivity changesup to 50% have been shown to occur at temper-atures below 100³C [49]. We strongly recommendthat temperatures above ambient should beavoided at any stage during sample preparationfor IR-OSL dating of feldspar separates.

It is not yet known whether the change in trap-ping probability in feldspar as a consequence ofheating is common. In a preliminary investiga-tion, the correction factor was determined forthree feldspar separates from Denmark (samples989201^989203 [50]) and one feldspar separatefrom New Zealand (sample GDNZ6 [8]). The cor-rection factors obtained for a 10 s, 290³C preheatwere all less than 1.1, indicating that the e¡ectdoes not produce an error greater than 10% forthese samples. For the samples from Denmark,identical results were obtained from the OSL dat-ing of the quartz fraction and IR-OSL dating ofthe feldspar separates (both used SAR procedures

Table 5Feldspar IR-OSL results corrected for sensitivity changes

Sample Correction factorSARa

Feldspar IR-OSL equivalent dose Feldspar IR-OSL age Quartz OSLage

(Gy) (ka) (ka)

SARb SARAc CorrectedSARd

SARb SARAc CorrectedSARd

Winssen 0.94 þ 0.04 2.9 þ 0.2 4.2 þ 2.1 2.7 þ 0.2 1.52 þ 0.14 2.3 þ 1.1 1.44 þ 0.14 0.92 þ 0.10Rumpt I-3 0.97 þ 0.02 2.8 þ 0.4 3.6 þ 0.6 2.7 þ 0.4 1.5 þ 0.2 1.9 þ 0.3 1.4 þ 0.2 1.23 þ 0.10d

Rumpt IV-2 1.217 þ 0.014 4.2 þ 0.3 6.9 þ 0.6 5.1 þ 0.4 1.47 þ 0.13 2.4 þ 0.2 1.8 þ 0.2 1.75 þ 0.10Schelluinen II-3 1.16 þ 0.02 7.4 þ 0.5 11.2 þ 2.2 8.5 þ 0.6 3.8 þ 0.3 5.8 þ 1.2 4.4 þ 0.4 5.1 þ 0.4Schelluinen II-6 1.13 þ 0.08 7.6 þ 0.4 9.5 þ 0.3 8.6 þ 0.8 4.0 þ 0.3 5.0 þ 0.3 4.5 þ 0.5 6.1 þ 0.5Elden 1.20 þ 0.06 16.8 þ 1.2 ^ 20.2 þ 1.7 7.9 þ 0.6 ^ 9.5 þ 0.9 13.3 þ 0.8Leidschendam I 1.22 þ 0.08 46.9 þ 1.8 52 þ 6 57 þ 4 28 þ 2 31 þ 4 34 þ 3 48 þ 4Leidschendam II 1.34 þ 0.08 47.6 þ 1.5 ^ 64 þ 4 32 þ 2 ^ 43 þ 4 55 þ 6Leidschendam III 1.32 þ 0.09 59 þ 4 ^ 78 þ 7 44 þ 4 ^ 58 þ 6 82 þ 9Leidschendam IV 1.41 þ 0.13 71 þ 5 ^ 99 þ 11 38 þ 3 ^ 53 þ 7 71 þ 6Leidschendam V 1.21 þ 0.13 52 þ 3 ^ 63 þ 8 35 þ 3 ^ 42 þ 6 61 þ 5Leidschendam VI 1.43 þ 0.04 65 þ 3 ^ 93 þ 5 38 þ 3 ^ 55 þ 4 58 þ 4Leidschendam VII 1.324 þ 0.007 146 þ 17 ^ 193 þ 23 100 þ 15 ^ 136 þ 18 120 þ 9Leidschendam VIII 1.34 þ 0.07 191 þ 13 ^ 256 þ 22 140 þ 16 ^ 194 þ 20 158 þ 13Leidschendam IX 1.36 þ 0.06 96 þ 6 137 þ 18 130 þ 10 71 þ 6 105 þ 8 96 þ 9 145 þ 16Leidschendam X 1.40 þ 0.05 112 þ 4 ^ 156 þ 8 78 þ 5 ^ 109 þ 8 180 þ 28a Correction needed to account for sensitivity changes due to preheating. Details are given in the main text.b The SAR protocol [28] was used for equivalent-dose estimation.c The SARA protocol [48] was used for equivalent-dose estimation.d As derived using the correction factors given in the ¢rst column.

EPSL 6028 4-12-01


for equivalent-dose determination [50]). Althoughit is not clear how widespread this problem is, werecommend that testing for the phenomenonshould be routine practice when single-aliquotIR-OSL dating is applied to feldspar separates.

It should be stressed that underestimation inthe luminescence age obtained on feldspar sepa-rates has also been found when multiple-aliquotprocedures were used [9]. Moreover, correctingfor changes in trapping probability or applicationof multiple-aliquot techniques does not eliminatethe age underestimation for our samples. This in-dicates that there must be additional reasons forthe underestimation. A possible candidate wouldbe incomplete removal of unstable charge fromlaboratory irradiation. Incorporation of the cor-rection factor for di¡erent preheats, as shown inFig. 4, suggests that a true preheat plateau doesnot exist for this sample; after correction forchanges in trapping probability the `plateau' risescontinuously. This indicates that unstable chargeis present up to temperatures where almost alltrapped charge is removed (10 s preheats 325³C). The apparent plateau shown for theuncorrected data (Fig. 4) appears to be a conse-quence of two competing phenomena: incompleteremoval of unstable charge giving a rising trendwith temperature, and changes in trapping prob-ability causing a decreasing trend.

6. Conclusions

OSL ages, obtained using the SAR protocol onquartz from £uvial channel deposits in the Rhine^Meuse Delta, are in excellent agreement with tightindependent age control for the range of 1^13 ka.Thermal transfer was shown to result in a smalloverestimation of age, but this unwanted e¡ectwas largely avoided by using a less stringent pre-heat regime. Our results con¢rm the applicabilityof quartz OSL dating to establish absolute chro-nologies for late Quaternary sedimentary recordsin general, and £uvial records in particular.

Single-aliquot IR-OSL dating of feldspar sepa-rates proved to be less successful. The IR-OSLage of the feldspar samples is underestimated byup to 50% in comparison with independent age

control (up to 13 ka), and quartz OSL datingresults (up to 200 ka). We show that part of thisage underestimation is caused by changes in thecharge trapping probability as a consequence ofheating of the sample during single-aliquot proce-dures. This problem can be circumvented by usingthe SARA protocol, or by determining a sample-dependent correction factor. Both procedures pro-duce results that are in better agreement with theindependent age control, but they only partlysolve the underestimation problem. Clearly, ourresults indicate that previously established lumi-nescence chronologies based on coarse-grain feld-spar may need re-evaluation.

Considering the problems encountered in theIR-OSL dating of feldspar, we suggest that quartzis the mineral of choice for OSL dating of thesedeposits, and probably of late Quaternary sedi-ments in general. Nevertheless, it is importantthat the problems with coarse-grain feldspar dat-ing continue to be investigated, in view of thepotential of feldspars to extend luminescence dat-ing to much longer time scales than quartz.

Acknowledgements

This is a contribution to the NEESDI (Nether-lands Environmental Earth System Dynamics Ini-tiative) program. J.W. is grateful for additionalfunding received from the Netherlands Organiza-tion for Scienti¢c Research (NWO), which al-lowed him to visit the Luminescence Laboratoryof the University of Wales in Aberystwyth, UK,and the Nordic Laboratory for LuminescenceDating, Aarhus University, Denmark, and tocarry out the OSL measurements at those facili-ties. Thanks are due to the Netherlands Instituteof Applied Geoscience (NITG-TNO) for drillingand processing of the Leidschendam core. Lor-raine Morrison (University of Wales Aberyst-wyth) kindly etched the quartz grains of the Ho-locene samples; Mette Adrian and Anne SÖrensen(Nordic Laboratory for Luminescence Dating)did the same for the other samples and helpedwith the Q-spectroscopy. The manuscript wasgreatly improved following thorough reviews byAnn Wintle, Michel Lamothe and an anonymous

EPSL 6028 4-12-01


reviewer. We greatly appreciate their interest andconstructive comments.[AH]

References

[1] M.J. Aitken, An Introduction to Optical Dating, OxfordUniversity Press, Oxford, 1998, 267 pp.

[2] A.S. Murray, J.M. Olley, G.G. Caitcheon, Measurementof equivalent doses in quartz from contemporary water-lain sediments using optically stimulated luminescence,Quat. Sci. Rev. 14 (1995) 365^371.

[3] A.S. Murray, Developments in optically stimulated lumi-nescence and photo-transferred thermoluminescence dat-ing of young sediments: application to a 2000-year se-quence of £ood deposits, Geochim. Cosmochim. Acta60 (1996) 565^576.

[4] J.M. Olley, G.G. Caitcheon, A.S. Murray, The distribu-tion of apparent dose as determined by optically stimu-lated luminescence in small aliquots of £uvial quartz: im-plications for dating young sediments, Quat. Sci. Rev. 17(1998) 1033^1040.

[5] S. Stokes, H.E. Bray, M.D. Blum, Optical resetting inlarge drainage basins; tests of zeroing assumptions usingsingle-aliquot procedures, Quat. Sci. Rev. 20 (2001) 879^885.

[6] S. Stokes, D.R. Gaylord, Optical dating of Holocene dunesands in the Ferris Dune Field, Wyoming, Quat. Res. 39(1993) 274^281.

[7] A.S. Murray, L.B. Clemmensen, Luminescence dating ofHolocene aeolian sand movement, Thy, Denmark, Quat.Sci. Rev. 20 (2001) 751^754.

[8] G.A.T. Duller, Luminescence dating using feldspars: Atest case from southern North Island, New Zealand,Quat. Sci. Rev. 13 (1994) 423^427.

[9] U. Radtke, A. Janotta, A. Hilgers, A.S. Murray, Thepotential of OSL and TL for dating Lateglacial and Hol-ocene dune sands tested with independent age control ofthe Laacher See tephra (12 880 a) at the Section `Mainz-Gonsenheim', Quat. Sci. Rev. 20 (2001) 719^724.

[10] T.E. To«rnqvist, Holocene alternation of meandering andanastomosing £uvial systems in the Rhine^Meuse Delta(central Netherlands) controlled by sea-level rise and sub-soil erodibility, J. Sediment. Petrol. 63 (1993) 683^693.

[11] H.J.A. Berendsen, E. Stouthamer, Late Weichselian andHolocene palaeogeography of the Rhine^Meuse delta,The Netherlands, Palaeogeogr. Palaeoclimatol. Palaeole-col. 161 (2000) 311^335.

[12] H. Middelkoop, Embanked £oodplains in the Nether-lands; Geomorphological evolution over various timescales, Neth. Geogr. Stud. 224 (1997) 1^341.

[13] T.E. To«rnqvist, G.J. Van Dijk, Optimizing sampling strat-egy for radiocarbon dating of Holocene £uvial systems ina vertically aggrading setting, Boreas 22 (1993) 129^145.

[14] T.E. To«rnqvist, A.F.M. de Jong, W.A. Oosterbaan, K.

van der Borg, Accurate dating of organic deposits byAMS 14C measurement of macrofossils, Radiocarbon 34(1992) 566^577.

[15] A. Verbraeck, Toelichtingen bij de Geologische Kaart vanNederland 1:50.000. Blad Tiel West (39W) en Blad TielOost (39O), Rijks Geologische Dienst, Haarlem, 1984, 355pp.

[16] M. Friedrich, B. Kromer, M. Spurk, J. Hofmann, K.L.Kaiser, Paleo-environment and radiocarbon calibration asderived from Lateglacial/Early Holocene tree-ring chro-nologies, Quat. Int. 61 (1999) 27^39.

[17] H.U. Schmincke, C. Park, E. Harms, Evolution and en-vironmental impacts of the eruption of Laacher See Vol-cano (Germany) 12.900 a BP, Quat. Int. 61 (1999) 61^72.

[18] J. van der Plicht, The Groningen radiocarbon calibrationprogram, Radiocarbon 35 (1993) 231^237.

[19] T.E. To«rnqvist, M.F.P. Bierkens, How smooth shouldcurves be for calibrating radiocarbon ages?, Radiocarbon36 (1994) 11^26.

[20] T.E. To«rnqvist, J. Wallinga, A.S. Murray, H. de Wolf, P.Cleveringa, W. de Gans, Response of the Rhine^Meusesystem (west^central Netherlands) to the last Quaternaryglacio-eustatic cycles: a ¢rst assessment, Glob. Planet.Change 27 (2000) 89^111.

[21] J. Wallinga, J. van der Staay, Sampling in waterloggedsands with a simple hand-operated corer, Ancient TL 17(1999) 59^61.

[22] E. Oele, W. Apon, M.M. Fischer, R. Hoogendoorn, C.S.Mesdag, E.F.J. de Mulder, B. Overzee, A. Seso«ren, W.E.Westerho¡, Surveying The Netherlands: sampling tech-niques, maps and their applications, in: M.W. van denBerg, R. Felix (Eds.), Special Issue in the Honour ofJ.D. de Jong, Geol. Mijnb. 62 (1983) 355^372.

[23] L. BÖtter-Jensen, E. Bulur, G.A.T. Duller, A.S. Murray,Advances in luminescence instrument systems, Radiat.Meas. 32 (2000) 523^528.

[24] A.S. Murray, A.G. Wintle, Luminescence dating of quartzusing an improved single-aliquot regenerative-dose proto-col, Radiat. Meas. 32 (2000) 57^73.

[25] E.J. Rhodes, Observations of thermal transfer OSL sig-nals in glacigenic quartz, Radiat. Meas. 32 (2000) 595^602.

[26] G.A.T. Duller, Equivalent dose determination using singlealiquots, Nuclear Tracks Radiat. Meas. 18 (1991) 371^378.

[27] G.A.T. Duller, Luminescence dating using single aliquotsmethods and applications, Radiat. Meas. 24 (1995) 217^226.

[28] J. Wallinga, A.S. Murray, A.G. Wintle, The single-aliquotregenerative-dose (SAR) protocol applied to coarse-grainfeldspar, Radiat. Meas. 32 (2000) 529^533.

[29] A.S. Murray, R. Marten, A. Johnston, P. Marten, Anal-ysis for naturally occurring radionuclides at environmen-tal concentrations by gamma spectrometry, J. Radioanal.Nuclear Chem. 115 (1987) 263^288.

[30] J.M. Olley, A.S. Murray, R.G. Roberts, The e¡ects ofdisequilibria in the uranium and thorium decay chains

EPSL 6028 4-12-01


on burial dose rates in £uvial sediments, Quat. Sci. Rev.15 (1996) 751^760.

[31] V. Mejdahl, Thermoluminescence dating beta dose attenu-ation in quartz grains, Archaeometry 21 (1979) 61^72.

[32] J.R. Prescott, J.T. Hutton, Cosmic ray contributions todose rates for luminescence and ESR dating large depthsand long-term time variations, Radiation Measurements23 (1994) 497^500.

[33] V. Mejdahl, Internal radioactivity in quartz and feldspargrains, Ancient TL 5 (1987) 10^17.

[34] J. Rees-Jones, M.S. Tite, Optical dating results for Britisharchaeological sediments, Archaeometry 39 (1997) 177^188.

[35] H.J.T. Weerts, Complex con¢ning layers; architecture andhydraulic properties of Holocene and Late Weichseliandeposits in the £uvial Rhine^Meuse delta, The Nether-lands, Neth. Geogr. Stud. 213 (1996) 1^187.

[36] D.W. Zimmerman, Thermoluminescent dating using ¢negrains from pottery, Archaeometry 13 (1971) 29^52.

[37] J. Wallinga, G.A.T. Duller, The e¡ect of optical absorp-tion on the infrared stimulated luminescence age obtainedon coarse-grain feldspar, Quat. Sci. Rev. 19 (2000) 1035^1042.

[38] M.J. Aitken, Thermoluminescence Dating, AcademicPress, London, 1985, 359 pp.

[39] A.S. Murray, J.M. Olley, Precision and accuracy in theoptically stimulated luminescence dating of sedimentaryquartz: a status review, Geochronometria, submitted.

[40] L. BÖtter-Jensen, V. Mejdahl, Determination of potassi-um in feldspars by beta counting using a GM multi-counter system, Nuclear Tracks Radiat. Meas. 10 (1985)663^666.

[41] S.H. Li, Optical dating insu¤ciently bleached sediments,Radiat. Meas. 23 (1994) 563^567.

[42] A.G. Wintle, Anomalous fading of thermoluminescence inmineral samples, Nature 245 (1973) 143^144.

[43] N.A. Spooner, The anomalous fading of infrared-stimu-lated luminescence from feldspars, Radiat. Meas. 23(1994) 625^632.

[44] M. Lamothe, M. Auclair, A solution to anomalous fadingand age shortfalls in optical dating of feldspar minerals,Earth Planet. Sci. Lett. 171 (1999) 319^323.

[45] D.J. Huntley, M. Lamothe, Ubiquity of anomalous fadingin K-feldspars, and the measurement and correction for itin optical dating, Can. J. Earth Sci. 38 (2001) 1093^1106.

[46] A.S. Murray, R.G. Roberts, Measurement of equivalentdose in quartz using a regenerative-dose single-aliquotprotocol, Radiat. Meas. 29 (1998) 503^515.

[47] J. Wallinga, A.S. Murray, G.A.T. Duller, Underestima-tion of equivalent dose in single-aliquot optical dating offeldspar caused by pre-heating, Radiat. Meas. 32 (2000)691^695.

[48] V. Mejdahl, L. BÖtter-Jensen, Luminescence dating of ar-chaeological materials using a new technique based onsingle aliquot measurements, Quat. Sci. Rev. 7 (1994)551^554.

[49] T. Trautmann, M.R. Krbetschek, A. Dietrich, W. Stolz,The basic principle of radioluminescence dating and alocalized transition model, Radiat. Meas. 32 (2000) 487^492.

[50] K. Strickertsson, A.S. Murray, H. Lykke-Andersen, Opti-cally stimulated luminescence dates for Late Pleistocenesediments from Stensn×s, Northern Jutland, Denmark,Quat. Sci. Rev. 20 (2001) 755^759.

EPSL 6028 4-12-01


testing optically stimulated luminescence dating of sand ...tor/documents/epsl2001.pdf19098,...

Documents