Electron spin resonance (ESR) dating of Quaternary materials

ESR dating has become an efficient tool in earth sciences for geochronological studies on different kinds of littoral deposits (coral reefs terraces, beach ridge systems, aeolianites) during the last ten years. Improvements in annual dose rate (D’) estimation and the newly developed approach for equivalent dose (DE) determination (DE -Dmax plot procedure) increase the precision of ESR dating of Holocene and Pleistocene corals as well as marine and terrestrial mollusc shells. This is strongly supported by the comparison of ESR dating results with other numeric dating methods such as radiocarbon and TIMS Uranium series analysis (TIMS 230Th/234U). The latter is the main focus of this paper. The uncertainties associated in ESR dating of Holocene corals coincide with the variability of 14C ages caused by the marine reservoir effect. The dating of Pleistocene corals permits the differentiation between the main marine isotope stages (MIS) 5, 7, 9, 11 and 13 as well as between sub-stages 5e3/2 and 5e1, 5c, and 5a2 and 5a1. The average error range when dating corals is between 5 to 8%. Furthermore, ESR dating of marine and terrestrial mollusc shells has yielded some promising results and permits the differentiation between the interglacial MIS 1, 5, 7 and 9 with an average dating error range of 10 to 15%. ESR dating of quartz is another promising dating technique for Quaternary and even Neogene geological formations. The presence of quartz in volcanic rocks, tephra, fault gouge and sediments (heated or unheated) allows determining the last time of heating, fault movement or sunlight exposure. Although challenged by several experimental issues, ESR dating of quartz is often the only method able to produce numerical ages for older formations. ESR has also been applied to a wide variety of other materials such as foraminifera, speleothems, travertines, calcretes and tooth enamel. The most common and reliable application is the ESR dating of mammal teeth, which becomes in conjunction with laser ablation U-series dating, an important method for determining the age of archaeological sites beyond the time range of the 14C dating method back to about 200 to 300 ka. [Elektronen Spin Resonanz (ESR)-Datierung quartärer Materialien] Kurzfassung: ESR hat sich im letzten Jahrzehnt bei der Datierung verschiedenster littoraler Ablagerungen (Korallenriffterassen, Strandwallsysteme, Dünen) als effizientes Datierungswerkzeug etabliert. Verbesserungen in der Bestimmung der jährlichen Dosisleistung (D’) und ein neu entwickelter Ansatz zur Bestimmung der Äquivalent Dosis (DE – Dmax Verfahren) haben die Präzision der ESR-Datierung sowohl an holozänen und pleistozänen Korallen als auch an marinen und terrestrischen Molluskenschalen verbessert. Dies wurde durch den Vergleich mit anderen numerischen Datierungsverfahren wie Radiokohlenstoff und TIMS-Uranserien-Analyse (TIMS 230Th/234U) unterstützt. Der Vergleich mit letzterer Methode steht im Fokus dieses Artikels. Die mit der ESR-Methode verbundenen Ungenauigkeiten bei der Datierung holozäner Korallen liegt in der Größenordnung der Variabilität von 14C-Altern, die durch den marinen Reservoireffekt bedingt ist. Die Datierung pleistozäner Korallen erlaubt die Differenzierung der wichtigen marinen Isotopenstadien (MIS) 5, *Addresses of authors: G. Schellmann, Universität Bamberg, Lehrstuhl Geographie II – Physische Geographie, Am Kranen 1, D-96045 Bamberg. E-Mail: gerhard.schellmann@ggeo.uni-bamberg.de; K. Beerten, Universität zu Köln, Geographisches Institut, Albertus-Magnus-Platz, D-50923 Köln. E-Mail: kbeerten@unikoeln.de; U. Radtke, Universität zu Köln, Geographisches Institut, Albertus-Magnus-Platz, D-50923 Köln. E-Mail: u.radtke@uni-koeln.de eschweizerbartxxx 7, 9,11 und 13 sowie der Untereinheiten 5e3/2 und 5e1, 5c und 5a1 und 5a2. Der durchschnittliche Fehler bei der Datierung von Korallen liegt zwischen 5 bis 8%. Weiterhin hat die Datierung mariner und terrestrischer Mollusken mittels ESR viel versprechende Resultate geliefert, die eine Differenzierung der Interglaziale MIS 1, 5, 7 und 9 ermöglichen, bei einem Fehler von 10-15%. Die ESR Datierung von Quarz ist eine weitere viel versprechende Datierungstechnik für quartäre und sogar neogene geologische Formationen. Das Vorkommen von Quarz in vulkanischen Gesteinen, Tephren, Störungen und Sedimenten (thermisch beeinflusst und unbeeinflusst) ermöglicht die zeitliche Bestimmung des letzten Zeitpunkts vor der Erhitzung, Störung oder der Aussetzung von Sonnenlicht. Obwohl durch einige experimentelle Ergebnisse angezweifelt, ist die ESR Datierung von Quarz die einzige Möglichkeit Altersdaten älterer Ablagerungen zu liefern. ESR wurde auch bei einer Vielzahl anderer Materialien angewendet, wie zum Beispiel Foraminiferen, Speleothemen, Travertinen, Kalkkrusten und Zahnschmelz. Die gebräuchlichste und zuverlässigste Anwendung ist die ESR-Datierung von Mammutstoßzähnen. Im Zusammenspiel mit der Laser-Ablation Uranserien Datierung ist ESR eine wichtige Methode zur Altersbestimmung archäologischer Fundstätten jenseits der Bestimmungsgrenzen der Radiokohlenstoffmethode bis in den Bereich von 200 bis 300 ka.


Introduction
Although IKEYA (1975) introduced Electron Spin Resonance Spectroscopy (ESR) for dating stalagmites more than 30 years ago, the full potential of this relatively young method is still not fully utilised. Since then, the quality of ESR-spectrometers and the understanding of the structure and behaviour of the ESR signal used for dating have been significantly improved. Detailed overviews about the ESR dating method have been provided by IKEYA (1993), GRÜN (1989aGRÜN ( , 1989bGRÜN ( , 2007, RADTKE (1989), JONAS (1997) and RINK (1997). This method is used for a wide variety of materials with most reliable applications on corals, mollusc shells, quartz, foraminifera, speleothems and teeth. In this text, the potential and present restrictions of ESR dating of Quaternary coral, mollusc shells and quartz is illustrated for selected sites and a short review is presented about the relevance of ESR for dating other carbonates (foraminifera, speleothems) and tooth enamel. For further details about problems and further applications of ESR dating including more literature, see e.g. GRÜN (2007), BLACKWELL (2006) and RINK (1997). Regardless of the recent methodological improvements in ESR and its frequent use for dating corals, mollusc shells, quartz and Electron spin resonance (ESR) dating of Quaternary materials 151 teeth, there is still a huge potential for further development of the method for these and other materials (e.g. SKINNER 2000). Similar to all other methods of age determination, ESR dating is confronted by specific methodological problems that cannot be considered in error calculation. This means, as applies for most analytic methods, that high precision does not automatically guarantee the accuracy of an age estimate. The latter is influenced by different geological factors that cannot be quantified and are hence not part of the age and error calculation. In ESR dating, these are mainly diagenetic alterations of the dated material, incomplete resetting of the ESR signal and up-take or loss of radioactive elements (U, Th, K) ( Fig. 1). With regard to these problems, proving the reliability of new dating approaches by independent age control is essential. In the context of the ESR method, this is possible using radiocarbon or g = 2.0007 GERHARD SCHELLMANN, KOEN BEERTEN AND ULRICH RADTKE 152 eschweizerbartxxx the mineral itself (D' int. , internal dose rate: i.e. Uranium and daughter isotopes, to some extent additionally Th and K in quartz) as well as radiation from the surrounding sediment (D' ext. , external dose rate mainly from Uranium and Thorium decay chains and Potassium). Such ionising radiation causes the activation of electrons to an excited energy level. Charge defects in the crystal lattice, so-called traps, capture part of the excited electrons in the band gap. A detailed description of the underlying physical processes is provided by e.g. GRÜN (1989aGRÜN ( , 1989bGRÜN ( , 2007, JONAS (1997( ) und RINK (1997. The amplitude of the ESR signal represents the amount of unpaired electrons at lattice defects (traps). Each material investigated has a characteristic ESR spectrum that may consist of one or several single signals, but not all the individual signals are suitable for dating. Suitable are only ESR signals that are both sensitive to radiation and thermally stable at the prevailing temperature that occurred during deposition and burial. The ESR signal at g = 2.0007 ( Fig.  1) is most suitable for dating aragonitic mollusc shells and corals (e.g. , WALTHER et al. 1992, SCHELLMANN & RADTKE 1999. A basic principle in dating biogenic material is that the ESR signal starts to increase after the shell, tooth enamel or coral has been formed. For ESR dating of quartz, a prerequisite is the existence of a resetting mechanism in nature to zero the ESR signal (and thus the geological clock). In order for the method to be reliable, any pre-existing ESR signals must be erased prior to the event to be dated (Fig. 2). The zeroing process responsible for resetting the geological clock is dependent on the geological context of the quartz mineral. The specific behaviour of ESR centres in quartz allows to determine the moment of the last heating (volcanic rocks, tephra, heated sediments), the moment of the last fault movement (quartz in fault gouges exposed to shearing) and the last exposure to sunlight (sedimentary quartz). The accuracy of the ESR age is dependent on the completeness of the zeroing process. Incomplete signal zeroing will inevitably lead to age Electron spin resonance (ESR) dating of Quaternary materials 153 eschweizerbartxxx and thus useful for dating. These are impurities related to Al-, Ti-and Ge-centres and the intrinsic E'-and OHC-centres (RINK 1997). An example of an ESR spectrum of quartz is shown in Fig. 3. Although the dosimetric properties of these centres differ significantly, they all show a relatively slow rise of the ESR signal with accumulating doses (Fig. 4). This specific behaviour of ESR centres permits to extend the dating range up to 1 Ma and more, i.e., beyond the range of current luminescence dating techniques. Any ESR age is calculated by dividing the dose accumulated with time (palaeodose, D E ) by the dose rate ( Fig. 1). As the number of defects and the rate of dose absorption are sample specific, the equivalent of the palaeodose is determined by a so-called additive dose response curve. For this, the sample is irradiated using artificial bor g-sources and an individual dose response curve is constructed for each sample. By extrapolation on the x-axis, D E is calculated (Fig.  1). The equivalent dose of quartz can also be determined using the regenerative dose method, owing to the regeneration characteristics of the paramagnetic defects in the mineral. Following thermal annealing (heating) or optical bleaching, the ESR signal is regenerated with an artificial radioactive source, and the dose is estimated by interpolation of the natural ESR intensity (Fig. 2).

Dose rate determination
The cosmic dose rate is related to the depth of the sample below surface, latitude and elevation. Further corrections to burial depth may be needed where nearby obstructions or drop-offs in the land surface occur: these are not accounted for by the tables of PRESCOTT & HUTTON (1994). The radioactivity of the surrounding material is either measured in the field using a portable gamma spectrometer or calculated via the determination of the concentration of Potassium, Thorium and . The data points are fitted to a rising exponential (GRÜN 1989). Positive values refer to the artificially added doses, while negative values refer to the past irradiation dose. Assuming a dose rate of about 1 Gy ka -1 , which is not uncommon in sedimentary contexts, the dose range covered in this example would equal more than 2 million years.
Abb. 4: ESR-Dosisaufbaukurve für das Al-Zentrum gemessen an einem äolischen Sediment (NWB1, Murray Basin, Australien, unveröffentlichte Daten). Die Datenpunkte sind an einen steigenden Exponenten angepasst (GRÜN 1989 For the calculation of ESR ages of corals, an early Uranium up-take and hence a relatively high radiation level from the very beginning can be assumed. This is of importance as ESR ages that are calculated based on an early Uranium up-take model will result in much lower age estimates compared to ages calculated based on the assumption of a linear Uranium uptake. In contrast to this, modern mollusc shells have extremely low Uranium contents of 0.1 to 0.2 ppm (max. 0.7 ppm). Only after an age of more than 2500 yr, do mollusc shells show significantly higher contents of more than 2 ppm Uranium ( Fig. 6) and reach levels frequently observed in Pleistocene fossils. This implies that mollusc shells absorb most of the Uranium post mortem. However, this delayed up-take of Uranium is negligible for age calcu-lation when dating Pleistocene mollusc shells. For Early to Mid Holocene mollusc shells, which have Uranium contents higher than 0.5 ppm, a delayed up-take of Uranium has to be considered when calculating the internal dose rate. As the exact post-mortem up-take of Uranium cannot be reconstructed, the true age will be between the ESR ages estimates calcu-   (1997) and RINK (1997). Depending on the contribution of internal dose (= Uranium content of mollusc shell) to the total dose rate, differences in age between the two models will be up to 1000 yr (and more) for samples having a high Uranium content (>5 ppm). For Uranium contents below 0.6 ppm, both models usually result in age estimates consistent within error.    RADTKE (2007). In addition to the Uranium content of radiocarbon dated mollusc shells, values measured for in situ bivalve shells that were collected from the same horizons are included.
Abb. 6: Uran-Gehalte (ppm) moderner, holozäner und pleistozäner Muschelschalen (aus SCHELLMANN & RADTKE 2007). Zusätzlich zu den Uran-Gehalten der 14 C-datierten Muschelschalen, sind auch die Werte von den in situ im gleichen Horizont gefundenen bivalven Schalen angegeben. eschweizerbartxxx corals since only cosmic radiation and internal Uranium are relevant for total dose rate ( Fig.  1). The problems of changing water content in the surroundings of the sample and erosive or accumulative processes that may affect cosmic dose calculation have only minor effects. Supplementary uncertainties in the dose rate for ESR dating of quartz are due to variations in water content and disequilibria in the decay chains of U and Th. Radioactive disequilibrium (i.e. a misbalance between parent and daughter nuclides) might occur if one (or more) of the Th-and/or U-chain members are lost or gained during burial, an effect that would be more pronounced in very permeable sediments. Especially, radionuclides of Rn (gas) and Ra (very leachable), which occur halfway in the U-and Th-chains, could induce disequilibrium. In the case of sediments, it is also possible that disequilibrium already existed at the time of deposition. In order to calculate an accurate dose rate, it is important to check whether any equilibrium exists and if so, whether this equilibrium has remained constant over time. However, in many dating studies, radioactive equilibrium is simply assumed. A further error source in the calculation of ESR ages is the only poorly known so-called alpha-efficiency, which is also known as k-factor (see JONAS 1997 for details). This value describes the efficiency of a-particles to induce ESR signals compared to other kinds of radiation. GRÜN (1985) and GRÜN & KAT-ZENBERGER (1994) experimentally determined by using a 241 Am a-source k-values between 0.07 and 0.10 for mollusc shells. For corals, GRÜN et al. (1992) and MALMBERG & RADTKE (2000) determined k-values between 0.05 and 0.07. According to LYONS (1987, cit. in RADTKE & GRÜN 1988 the actual alpha-efficiency of particles from the Uranium decay chain is 20-30 % higher then the values determined using a monoenergetic artificial a-source. Following GRÜN (2007GRÜN ( : 1509, the best k-values for molluscs are 0.07 ± 0.01, for corals 0.06 ± 0.02 and for tooth enamel 0.13 ± 0.02. The influence of uncertainties on the a-efficiency in quartz is re-duced because the internal alpha dose rate can usually be neglected (cf. supra). Furthermore, the influence of external alpha rays (several 10 µm) is erased by sample preparation techniques using HF-solutions. As such, in many cases the total dose rate to quartz can be simplified to the sum of the external beta and gamma dose rate and the cosmic dose rate.

ESR dating of aragonitic coral and marine and terrestrial mollusc shells
Besides the problems in dose rate determination mentioned above, ESR dating of mollusc shells and corals is associated with another uncertainty that has yet not been quantified. This problem is related to the amplitude of the ESR signal of most mollusc shells, and also rarely for corals, which does not show simple exponential growth to saturation resulting from artificial gamma irradiation. Instead, so-called inflexion points ( Fig. 1) have been observed at which signal growth increases suddenly. The physical nature of these inflexion points is yet only poorly understood. Most likely, it results from interference of the ESR dating signal with the so-called a-complex, which shows a relatively higher increase at higher doses and apparently individually disturbs the signal used for dating (KATZENBERGER & WILLEMS 1988. However, it is also possible that inflexion points are related to defects in the crystal lattice that are produced by gamma irradiation (GRÜN 1990). Attempts to eliminate inflexion points without alternating the ESR dating signal by changing the parameters of the ESR measurement or including thermal pre-treatments have not been successful so far (e.g. BRUMBY & YOSHIDA 1994a, HOFFMANN et al. 2001, MOLODKOV et al. 1998). Different preheat temperatures and durations can actually cause significant differences in D E values and consequently different ESR ages (SCHELL- MANN & RADTKE 2007). Discontinuous growth of the ESR signal can cause overestimation of calculated D E of more than 10 % when too high artificial doses are used for construction of the dose response curve (Fig. 1). For determination Electron spin resonance (ESR) dating of Quaternary materials of D E values, only the undisturbed, low-dose part of a dose response curve prior to reaching the first inflexion should be used. Only this part is dominated by the growth of the ESR dating signal and should reflect natural increase. This part of the growth curve can be described by a simple exponential saturation curve. The impact of inflexion points can be minimised by either using several dose points in the lower part of the dose response curve or by applying the standardised procedure of D E -D max -plots (DDP) ( Fig. 1; SCHELLMANN & RADTKE 1999. Nevertheless, inflexion points are probably the main reason for age scatter observed in isochronal samples within mollusc bearing sediment layers. Additive dose response curves of corals, on the other hand, rarely show pronounced inflexion points, presumably the reason why ESR dating of isochronal corals scatter much less than molluscs (see below). The D E values of all ESR ages mentioned here were determined using D E -D max plots. Prior to this, mollusc and coral samples were ground by hand and sieved to 125-250 µm. At least 20 aliquots with a weight of 0.2000 g were prepared and irradiated using a 60 Co-source (Centre of Nuclear Medicine, University of Düsseldorf); dose rate between 0.8 and 2.5 Gy min -1 ). The maximum irradiation dose was typically between two and three times D E . Typical parameters on the ESR spectrometer were 10 or 25 mW microwave power, 0.5 or 1.0-1.2 G modulation amplitude, 41.9 s scan-time, 40-50 G scan width and 5 to 40 scans. All D E values were determined using the programme "Fitsim" (version 1993) and ESR ages were calculated using the programmes "Data IV" (version 1990) and "Data V.6" (version 1999), written by Rainer Grün.

Comparing ESR, TIMS Th/U and radiocarbon dating results ages of mollusc shells from Holocene as well as Late and Middle Pleistocene littoral terraces of the Patagonian Atlantic coast
Since several early systematic studies (RADTKE et al. 1981, IKEYA & OMUHRA 1981, dating of molluscs is one of the most common applications of ESR dating and has been applied in several regions worldwide. Particularly typical for the example presented here, the Atlantic coast of Patagonia, is the phenomenon that the coarse littoral beach deposits frequently bear articulated mollusc shells (Photo 1). Such objects are very sensitive to movement and clearly indicate their in situ nature. This could be proven by radiocarbon dating of several bivalve mollusc shells found in one sediment layer (SCHELLMANN 1998, SCHELLMANN & RADTKE 2007 . 7). A similar picture is also revealed when comparing the whole data set of ESR and Th/U dating from different localities along the Patagonian Atlantic coast (Fig. 8). Only the ESR ages allow, with an error of 10-15 %, a geochronological differentiation between Last and Penultimate Interglacial littoral terraces. In cases were several datings are available, a differentiation of beach deposits belonging to Marine Isotope Stage (MIS) 9 is partly possible. For Holocene mollusc shells, radiocarbon dating can be an independent reliability control for ESR dating. Although such a comparison shows fairly good agreement and reproducibility, it is obvious that significant discrepancies do occur (Fig. 9). It should be kept in mind that articulated mollusc shells within a sediment layer certainly are of the same age. As the external dose rate within the sediment is equal for all samples and possible variations of internal dose rate cannot account for the observed age differences, the variation in age must be related to D E calculation and some not yet known properties of the ESR dating signal at g = 2.0007. We can conclude that ESR ages of Holocene and Pleistocene mollusc shells can scatter substantially and result in significantly too high age estimates. It is hence necessary to date several shells out of a sediment layer to establish a reliable chronological frame for marine terraces. The given accuracy of ESR dating of mollusc shells will, however, only allow correlating the littoral deposits with certain interglacials but not more precisely. are spread over several kilometres along the coast and are exposed in cliffs caused by littoral erosion. In some areas, such as along the coast at Nissi Beach (Fig. 10), the basal part of the aeolianite is visible just below present sea level. As the southern coast of Cyprus has only been weakly uplifted since the Last Interglacial (SCHELLMANN & KELLETAT 2001), the deposition of the aeolianite probably occurred during times when the sea level was near the coast and therefore only a few ten metres below its present position. During the Late Pleistocene, such sea level stands occurred during the later parts of MIS 5. Snail shells of Helix sp. have been used to confirm this geomorphologic interpretation of the age of the aeolianites by means of geochronology (Fig. 10) (Fig. 8). ESR dates of single shells from these marine deposits indicate an age of 130,000 to 137,000 yr. All together, the results from Cyprus underline the potential of ESR to date terrestrial mollusc shells.

ESR dating of Pleistocene corals from Barbados
Due to the fact that dose rate determination is relatively unproblematic, as already discussed above, the dating of aragonitic corals has a rather high potential. First test studies on ESR dating of corals were published in the late 1980s (IKEYA & OMUHRA 1983, RADTKE 1989, GRÜN et al. 1992) and since then, several methodological improvements and the development of more stabile and high-resolution ESR spectrometers have significantly increased the quality of the dating results. It is now not only possible to distinguish between the major periods of high sea level during the last 500,000 yrs but also to differentiate between sub-maxima, for example, during MIS 5 (e.g. SCHELLMANN age are on average not more than 3000 yr and hence within the error of ESR dating. Generally, TIMS U/Th of Late Pleistocene corals is considered to be a highly precise method with analytical errors of < 1 %. However, comparing the results of ESR and TIMS Th/U dating of Last Interglacial corals from Inch Marlowe Point (Fig. 12) and Batts Rock Bay (Fig. 13) reveals that the quality of TIMS U/Th is not better than that of ESR ). At Inch Marlowe Point, ESR ages are about 1000 to 8000 yr younger than TIMS U/Th regardless the actual age of the samples. The upper quartile value of all ESR ages is 73,100 yr and the median of all U/Th ages is 76,700 yr. Interestingly, the spread of individual ages (not considering error) for both methods is about 4000 yr, although all ages were produced on individuals from the same branch of corals and are hence most likely of the same age (grown within not more than a few hundred years). Hence, the high precision of the TIMS U/Th measurements (~ 1 %) apparently does not completely account for the observed scatter in ages of isochronal samples. In contrast, the observed scatter of ESR dating is explained by the relatively high analytical error (low precision) (5-8 %). Hence, when considering the whole data set, the accuracy of were systematically investigated (Fig. 11). At an elevation of 21 to 43 m above present sea level (asl.), three different coral reefs that formed during the transgression maximum of MIS 5e were identified. Three coral reefs at an elevation between 4 and 17 m asl. are correlated with the sub-maximum of MIS 5c and two further reef terraces at 2-3 m asl. are interpreted to represent sub-stages MIS 5a 1 and MIS 5a 2 ( The relatively large scatter of ESR ages determined for individual coral reef terraces is most likely caused by weak diagenetic alterations, which cause re-crystallisation and/or up-take of Uranium. The result is that some ESR ages underestimate the real age of the sample. As a consequence, the oldest ESR samples are more likely to represent the actual age of a coral reef and should hence be used for the chronological interpretation. When more than 20 ESR ages were available for a coral reef terrace, this was accounted for by using the 90 percentile value, and if less than 20 ages were available, then the upper quartile value was used to determine the mean age of the reef. Comparing the results of this approach with the median of U/Th dating shows a rather good concordance (      individual TIMS U/Th ages is not better than that of ESR dating. The best possible resolution of both dating methods is also nicely demonstrated for branches of corals sampled at Batts Rock Bay, west coast of Barbados, that were formed during the two sea level sub-maxima of MIS 5c und MIS 5a. According to this data set, both dating approaches are associated with a relatively high non-systematic spread of ages of a few thousand years.

ESR dating of Holocene corals from the Netherlands Antilles (Aruba, Bonaire, Curaçao)
To further confirm the accuracy of ESR dating, 21 radiocarbon-dated sub-modern, as well as Late and Middle Holocene, coral samples from Tsunami deposits found on Aruba, Bonaire and Curaçao (Netherlands Antilles) were dated by ESR (Fig. 14). The calibrated radiocarbon ages spread between 0 and 3644 years and the ESR ages were between 9 and 3653 yr (not considering the individual uncertainties of each measurement). Despite one sample that is considered an outlier, the ages determined for both methods are consistent within an error range of ca. 250 yr.
However, most ESR data shows a tendency towards slightly higher values in ages compared to radiocarbon, which were reservoir corrected assuming that ΔR = -49 (ca. 392 yr). It is likely that the marine reservoir effect did considerably change in the past and that the assumed ΔR-value is not representative for all samples. The calibrated radiocarbon ages may hence not automatically represent the "true" age of a sample.

ESR dating of heated quartz
As ESR signals in quartz usually decay when heated, heating mechanisms in nature may provide an appropriate means of resetting the geological clock, enabling one to establish the age of the (last) heating. For instance, xenolithic quartz in volcanic rocks in the Eifel region (Germany) was successfully dated with the Al-centre . In general, the ESR ages were in good agreement with independent age control (Ar-Ar) up to 500 ka (Fig. 15). Similarly, quartz-rich sediments, which were sufficiently heated by overlying lava flows, can be used to estimate the geological age of the lava flow. MIALLIER et al. (1994a) showed the potential of Ti-and Al-centres to date sediment baked by a Late Pleistocene lava flow by comparison with red TL and other independent age controls. However, dating attempts of heated quartz from a ~ 580 ka pumice ( 40 Ar/ 39 Ar) showed scattered ESR underestimates for these centres (MIALLIER et al. 1994b), leading to the conclusion that the dating range of ESR is below 500 ka for such materials.
ESR results for heated quartz were recently published by TOYODA et al. (2006). In this study, Al-and Ti-centres were used for ESR dating of Quaternary tephra from Japan, in comparison with red TL and other independent age controls (fission track, K/Ar). Although some degree of consistency could be observed for samples younger than 60 ka, the ESR results appeared very scattered, internally (inconsistencies between Al-and Ti-centres) as well as externally (inconsistencies between the ESR and TL data).
Reported ESR ages of heated quartz are often lower for the Al-centre, in comparison with the Ti-centre. This may be due to different thermal stabilities of Al-and Ti-centres (TOYODA & IKEYA 1994). Furthermore, high ambient temperatures such as those in geothermal areas may be sufficient to lower the natural ESR signal of volcanic quartz, eventually leading to age underestimates (TOYODA et al. 1995). RINK (1997)

ESR dating of strained quartz
ESR dating of faults is based on the principle of signal zeroing due to fault activity. Fault movements may strain quartz in fault gouge in such a way that the ESR signals are completely zeroed during faulting. Subsequently, the ESR signal starts to grow again, thus allowing the determination of the geological age of the last fault movement. ESR dating of fault gouge is of particular importance in seismic risk assessment. Generally, fault gouge is dated by the ESR plateau method (BUHAY et al. 1988). The methodology involves the use of several ESR centres (usually E'-, OHC-and Al-centres) from different quartz grain size fractions. The idea is that the smaller the grain size, the more the quartz is affected by strain and the more the geological clock has been reset (Fig. 16)

ESR dating of sedimentary quartz
Quartz grains that are exposed to natural sunlight sufficiently long will have their ESR signals reset to a certain extent during transport, prior to burial (Fig. 17). This important principle forms the basis of ESR dating of sediments. It is one of the most promising sub-disciplines in ESR dating. The specific age range of ESR dating of quartz, possibly up to several Ma, makes sediment dating a potentially very valuable tool, with applications in geology, geography, prehistory and palaeoanthropology. The specific issue in ESR dating of sediments is the extent to which the geological clock is reset to zero, or to a value, which can be determined in the laboratory. Resetting of the geological clock proceeds through sunlight bleaching of ESR centres during transport, prior to burial. Experimental bleaching curves are shown in Fig. 17 for various ESR centres and various artificial light sources. The Al-centre contains an unbleachable residual, whereas Ti-centres can be fully (or to a negligible level) reset.
Note that bleaching times are on the order of days to weeks.

Aeolian sediments
The most straightforward application of ESR dating of sedimentary quartz is that of aeolian sediments, because of the sunlight bleaching potential. Several studies report reasonable ESR ages for aeolian sand, based on the Alcentre, in the time frame up to several 100 ka (YOKOYAMA et al. 1985, TANAKA et al. 1995.
Since the Al-centre contains an important unbleachable residual with respect to natural sunlight radiation, it is important to mention that the ESR ages reported in these studies are corrected for this residual. The principle of this method is shown in Fig. 18. Recent investigations in ESR dating of aeolian sand using the Al-centre also include Chinese loess deposits (YIN et al. 2007 Fig. 20;LAURENT et al. 1998, Antoine et al. 2000. The same methodology, i.e. ESR of the Al-centre with subtraction of the unbleachable residual, has been used for other terrace staircases as well, with promising results (VOINCHET et al. 2004). The Al-centre even has the potential to date Plio-Miocene estuarine deposits, as was shown by LAURENT et al. (1998). In this study, a consistent ESR dating pattern for deposits from the Tiglian up to the Tortonian (around 8-9 Ma) was obtained. The quality of the ESR ages from these studies is guaranteed to some extent by bleaching experiments on recent fluvial deposits. LAURENT et al. (1998) and VOINCHET et al. (2003, 2007 showed that the Al-related ESR signal in quartz is at or very close to the unbleachable residual

Glacial sediments
Grinding of quartz grains may occur not only in a fault-related context, but also in some sedimentary environments. Together with sunlight bleaching, such grinding mechanisms could be an effective zeroing mechanism in ESR dating of glacially derived sediments. As such, the Ge-related impurity centre in quartz appears to generate promising results, e.g. in dating Pleistocene glaciations in Central Asia (ZHAO et al. 2006. Whereas the basis for this specific ESR approach was laid by methodological studies on the behaviour of Ge-centres with respect to sunlight bleaching and pressure (TANAKA et al. 1985, BUHAY et al. 1988, YE et al. 1998, some fundamental points remain unclear. For instance, although growing with small artificial irradiation doses, it appears very difficult to find a natural Ge signal in quartz, as has been pointed out already by RINK (1997). The implication of this observation could be that the Ge centre is not stable at average burial temperatures.

General remarks
Various methodological issues are common to all types of quartz, irrespective of the geological context. Generally, the additive dose method is used to determine the equivalent dose in quartz. However, sometimes this approach may lead to large uncertainties for samples with a relatively large equivalent dose. In such cases, it would be advantageous to use the regenerative dose method. For ESR centres in quartz, artificial heating may be the most efficient and comfortable way of annealing the signal prior to regenerative dosing. Although this mechanism is not the annealing mechanism in sedimentary environments, this approach is currently being tested for ESR dating of sediments using Ti-related centres (BEERTEN et al. 2006). A major concern here is the possible presence of sensitivity changes (i.e. an alteration of the radiation-sensitivity of the ESR-signal) due to the heating step. Less comfortable in experimental terms, but much more related to the zeroing mechanism in sedimentary environments, is optical bleaching of ESR centres prior to administering the regenerative dose. First results of this approach were recently published for Al-centres in loess (YIN et al. 2007). Unfortunately, the ESR dose response is often found to be problematic in various quartz samples. Sometimes, strong inflexion points are found in the dose response of the Ti-centre and the Al-centre. There are also observations on non-monotonic linear growth of Ti-and Ge-centres, indicating an ESR signal decrease with very high doses of several 1000 Gy (WODA & WAGNER, in press).

ESR dating of some other materials (speleothems, foraminifera, tooth enamel)
ESR dating has also been applied to a wide variety of other materials with great importance, especially in palaeoclimatological and archaeological contexts. The most common applications in this framework include speleothems, foraminifera, and tooth enamel. Many ESR papers have published combined ESR and U-series dating results of speleothems (stalagmites, stalactites), which generally show good agreement. A detailed review is given by RINK (1997: 990f.). Nevertheless, as stated by PIROUELLE et al. (2007), further research is still needed about the best analytical routine for obtaining correct D E values, which include systematic dating applications coupled with other (e.g. U-series) dating methods. However, as GRÜN (2007: 1512) stated, U series dating of speleothems and other secondary carbonates is much cheaper, faster and more accurate than ESR. Studies on ESR dating of planctonic foraminifera from a deep-sea sediment core by MUDELSEE et al. (1992) show good agreement with delta 18 O stratigraphy. The authors see the possibility to date planctonic foraminifera up to approx. 800 ka by using the ESR signal at g = 2.0036. More recently, HOFFMANN et al. (2001) stated that foraminifera could be dated up to 190 ka, just when using the CO 2 --signal at g = 2.0006. All in all, more studies on wellstratified deep sea cores in combination with independent age controls are needed to show GERHARD SCHELLMANN, KOEN BEERTEN AND ULRICH RADTKE the potential and the limits of ESR dating of foraminifera. Lately, ESR dating on human tooth enamel, in conjunction with laser ablation U-series dating, has become an important method for determining the age of human remains back to about 200 to 300 ka, beyond the time range of the 14 C dating method (e.g. GRÜN 2006GRÜN , 2007SORESSI et al. 2007). Most recently, GRÜN (2006) gives a detailed description of the ESR dating method itself in application on human teeth, which includes application examples and comparisons with other dating results ( 14 C, U-series, AAR). In addition, HAMEAU et al. (2007) see a good agreement of coupled ESR/Uranium-series dating of Middle Pleistocene Rhinocerus and Tapirus tooth enamel on East Java associated with lithic artefacts, which seems to provide a new geochronological framework of human occupation in South-East Asia. Details on ESR dating of teeth are summarized by GRÜN (2006) and RINK (1997). There are further materials, e.g. travertines, calcretes or bones, which have been the object of ESR dating. However, the ages are less reliable due to complex Uranium uptake histories, to problems of recrystallisations or to carbonate impurities, which often lead to age underestimations (e.g. RINK 1997; GRÜN 2007).

Summary and outlook
Due the methodological advances of the past decade, ESR dating has become an important geochronological method, especially for dating deposits such as coral reefs and beach ridges as well as quartz from different geological settings. The reliability of ESR dating of carbonates was proven by direct comparison with other dating methods such as radiocarbon and TIMS U/Th. ESR dating of Holocene corals shows accuracy similar to radiocarbon dating when all sources of uncertainty are fully considered. ESR dating of Pleistocene corals allows not only to distinguish between MIS 1, 5, 7, 9, 11 and 13 deposits, but also between sub-stages 5e 3/2 , 5e 1 , 5c, 5a 1 and 5a 2 . The average error of ESR dating of corals is about 5-8%. Late Pleistocene corals can be dated with a similar accuracy as with TIMS U/Th in this time range. The advantage of ESR, however, is that the upper dating limit for corals is probably above 500,000 yr. If diagenesis could be excluded, dating of coral with an age of several million years should be feasible from the physical point of view. The resolution of ESR dating of molluscs and gastropods is with 10-15 % much lower. Additionally, ages tend to scatter quite substantially, which causes the need to carry out several datings for the same stratigraphic unit to allow accurate geochronological interpretation. Usually it is then possible to correlate layers to certain interglacials periods such as MIS 1, 5, 7 and 9. The high scatter associated with ESR ages of molluscs and gastropods is most likely related to yet not identified problems in palaeodose determination, since problems in dose rate apparently do not account for the scatter. The examples shown in this paper demonstrate that ESR can be successfully applied to date quartz in various geological contexts. It should be clear that the large age range of ESR dating (up to several Ma), especially in relation to OSL dating (up to several 100 ka), underlines its important position next to other geochronological techniques. However, there are several issues that need further consideration. The accuracy with which the palaeodose can be determined is highly dependent on the degree of resetting of the ESR signal. Some examples show that incomplete resetting could undermine the reliability of the method. Therefore, it is important to incorporate criteria (external and internal) to evaluate the completeness of the zeroing process in nature, such as the plateau method for fault gouge and the use of multiple ESR centres and/or single grain measurements for sediment dating. Cross-checking of the results with independent age control is another crucial step in the development of a reliable ESR dating method. In the case of sediments, reference ages may be provided by luminescence dating (OSL, TL) for Late Pleistocene and in favourable circumstances even Middle Pleis-Electron spin resonance (ESR) dating of Quaternary materials tocene deposits. Similarly, ESR dating results of tephra layers could be evaluated against red TL measurements. The reliability of the palaeodose determination method, purely experimentally, is another concern in ESR dating of quartz. Results of the regenerative dose method and additive dose method should be compared very carefully and a proper annealing mechanism should be looked for in the regenerative dose method (UV-bleaching for sedimentary quartz; thermal annealing for heated quartz). At present, experimental techniques in ESR dating of quartz produce relatively large errors (usually around 15%). In part, this may be related to the dose determination method (i.e. the additive dose technique in the case of older samples), but it is important to verify if intrinsic properties of ESR defects could be the cause of such large errors (i.e. inflexion points). In general, it is believed that the regenerative dose technique may produce much more precise results if there are no sensitivity changes. There are even more materials, which have the potential for a dating by ESR, however the most reliable applications until now or in the near future seem to be coupled ESR/U-series dating of tooth enamel and perhaps speleothems. A great potential is inherited in ESR dating of foraminifera, which only needs more applications. In summary, it is clear that the development of the ESR method is far from being complete and it is important to carry out methodological investigations before the full potential of the method can be used.