For several years, thermoluminescence has been applied to the task of determining the dates of sedimentary deposition events. The method relies on the exposure of the sediment grains to sunlight at the time of their deposition to greatly reduce (or bleach) the TL signal. The most suitable types of sediment are therefore aeolian, low energy fluvial and lacustrine deposits, while rapidly deposited fluvial and sub-glacial sediments can be problematical.
Nature has provided us with a choice of TL signals from sedimentary material, and the dating specialist is free to select the signal that is best suited to the task. As measurement methods vary between laboratories, so do the properties of the TL that is observed. Therefore, the following notes should not be read as a review of sediment dating in general, but rather as a description of the technique employed at Quaternary TL Surveys and the reasons behind it.
4.1. Principles of TL Sediment Dating
In common with many crystalline materials, the minerals that compose sediments emit TL signals when subjected to progressively increasing temperature. One signal in particular is sensitive to light, and loses most of its intensity when exposed to illumination; the TL signal is then said to be bleached. When grains of sediment are transported prior to deposition, the exposure to daylight causes bleaching of the signal. After deposition, their latent TL slowly regenerates as the material, now shielded from light, responds to ambient levels of ionising radiation. Thus the present-day TL intensity of the sediment reflects both the environmental radiation level and the length of time since deposition. By measuring the TL and radiation intensities, the age of the sediment may be deduced.
The bleaching of the TL signal constitutes the initialising event without which TL dating is impossible. For the dating method to be reliable, there must be good reason to believe that the sediment grains were fully bleached before deposition. Sample selection is carried out with this requirement in mind, as explained in section 4.2. TL measurement procedures have an important part to play in selecting the TL signal that is most readily bleached. Use of this signal helps to ensure that the TL was at its minimum, or residual, level at the time of deposition. The method of palaeodose measurement is described in section 4.3.
The dating signal, in addition to being readily bleached, should also continue to grow when exposed to high doses of radiation, and should remain stable over long periods. Unfortunately, the rapidly bleached signal is not stable enough to allow the dating of sediments older than 150 ka (=150,000 years). In section 4.4 this age limitation is explained in more detail. Factors affecting the precision of sediment dates are described in section 4.5.
4.2. Selection of Sediment Samples for TL Dating
The primary consideration when selecting sediment samples for TL dating is whether the individual grains of the material received an adequate exposure to daylight at the time of deposition. Depending on the intensity of the daylight, exposures of between one day and one week are sufficient to reduce the intensity of the dating signal to its minimum level. Sample selection should aim to maximise the probability that a sufficient exposure occurred.
4.2.1 Suitable types of sediment
Aeolian material will certainly have received a long daylight exposure during its transportation. Among fluvial and lacustrine sediments, the fine-grained, low energy components are preferable, favouring material that was carried over long distance. Coarse, high energy fluvial sediments, by contrast, are associated with short transport times and muddy waters which daylight may not have penetrated. Gradually formed colluvial deposits are suitable for TL dating, but debris flows and other forms of mass transport are not. Soil formation can be dated since bioturbation causes the required bleaching.
In caves, aeolian deposits close to the entrance will be datable, but fluvial deposits in deeper locations may represent material which has been transported entirely in the dark. Glacially derived sediments present a variety of deposition conditions, but the prevalence of rapidly formed, high energy deposits and dark, sub-glacial situations means that TL application will often be problematical.
4.2.2 Requirements of the radiation dose rate measurement
Secondary to the consideration of TL bleaching, dose rate assessment imposes another set of conditions on sample selection. TL dating of sediments has an advantage over that of flint or calcite because there is often a wide choice of sampling locations. Samples should be collected at the same time as the environmental radiation survey so that the most suitable location for dose rate assessment can be chosen.
The purpose of the radiation survey is to measure the concentrations of the naturally occurring radioactive elements, uranium, thorium and potassium, in the sedimentary deposits which surround the sampling location. The measurements are performed with a portable gamma spectrometer which counts gamma rays emitted by the radio-isotopes, and identifies their energies (and hence, their emitter). Samples of the burial deposits are collected for supplementary laboratory analyses of their radioactivity, and for measurement of their water content.
Preferred sampling locations are those where the sedimentary deposits have large depth and are homogeneous. The gamma rays recorded during on-site dosimetry have a range of 0.3 m. Therefore, measurements performed at the centre of a strata of more than 0.6 m thickness will record radioactivity contained only within that level. Pebbles and larger objects within the deposit generally contain lower concentrations of radio-isotopes, and are large enough to influence gamma dose rates in their immediate vicinity. For this reason, it is desirable to sample homogeneous regions of the sediment where dose rates are uniform.
As explained above, sample collection is best carried out at the same time as the radiation dosimetry, and is therefore usually performed by the TL specialist. To avoid accidental bleaching of the TL, the samples must be shielded from daylight during collection. Instructions on how to do this will be supplied to fieldworkers if the need arises.
4.3. The Structure of Sediment TL
Thermoluminescence is observed as grains of sediment are progressively heated. This heating releases energy which is stored in the electronic configuration of the crystal, and which was previously imparted to the material by an exposure to radiation. Part of this energy appears in the form of light, and causes the TL emission. The available energy can remain stored in the crystal over long periods of time without significant loss, and therefore constitutes a long-term memory whereby the crystal remembers its past radiation dose. The energy is released from storage locations within the crystal (called traps) by the action of thermal energy supplied by the heating. As the temperature is increased, the available thermal energy also increases, and deeper traps are emptied of their contents. Thus, different types of traps are characterised by the temperature at which their associated TL emission occurs.
The graph of TL intensity vs temperature is called a glow curve. Figure 1 shows a typical glow curve (labelled N) recorded when fine grains of sediment were heated at a rate of 2°C per second. Curve N is the TL that the sediment possesses as a result of its exposure to natural radiation during burial, and is referred to as its natural TL. This glow curve is formed by the overlapping of two distinct TL signals which are associated with different trap types. The glow curve, N, can be analysed into its component signals, and these are shown in figure 1, labelled A and B; thus, A + B = N. Signal A has its maximum TL intensity at approximately 260°C, while signal B peaks at 340°C.
4.3.1 Selection of the TL dating signal
The 260°C signal is much more readily bleached than the peak at 340°C, and is therefore the preferred signal for dating applications. The precise temperature at which the 260°C signal has its maximum intensity is found to shift slightly depending on the age of the sediment. This is because the traps that store its latent TL lie at a range of depths. The shallower traps are slowly emptied by ambient thermal energies over the long burial period of the sediment. The older the sediment, the greater is the thermal loss of latent TL from these shallow traps. The temperature at which signal A starts to luminesce shows the extent to which TL has already been lost during the burial of the sediment. Thus, the rising edge of signal A occurs at a slightly higher temperature in old sediments than it does in younger samples.
The extent of the TL loss can be measured by comparing the natural TL of the sediment with that produced by an artificial irradiation in the laboratory. In the latter case, there is not a sufficient time lapse between irradiation and TL measurement to allow any significant thermal loss of signal intensity. The curve labelled T in figure 1 gives, on the right-hand axis, the percentage of natural TL which has survived intact at each temperature. When using TL for dating applications, it is clearly necessary to ensure that the selected TL has not suffered significant thermal loss; or to put it another way, that the crystal's memory of its past radiation exposure is unfaded.
Figure 1. Natural glow curve, N, of a sample of sediment analysed into its component signals, A and B (A + B = N). Curve T shows, on the right-hand scale, the extent of the loss of TL by thermal release at each temperature, while curve CA gives the proportion of signal A in the natural TL (CA = A ÷ N). As decribed in the text, the narrow temperature window, 290 - 300°C, provides the optimum selection of TL for dating purposes.
As stated above, the 260°C signal is the preferred one for dating because it is strongly reduced by an exposure to daylight, and therefore the quantity of TL in the sediment immediately after deposition is most likely to have been at the minimum, or residual, level. For this reason, it is important to select for the dating measurements as pure a sample of the 260°C signal as possible. In figure 1, the curve labelled CA gives, on the right-hand scale, the percentage of signal A in the natural TL at each temperature (CA = A ÷ N). It should be noted that the proportion of signal A to signal B varies between different sediments, reflecting the variability of their mineralogical compositions. However, for all samples, the selection of TL at ultra-violet wavelengths helps to suppress the unwanted signal B relative to signal A; this is achieved by placing a suitable filter in front of the light detector.
From the form of curve CA, it is evident that the selection of TL from a low temperature region maximises the proportion of signal A, and minimises the unwanted TL from signal B. However, as described above, the thermal loss curve, T, imposes a minimum temperature on the choice of TL, and a compromise therefore needs to be struck. For the example in figure 1, the optimum selection of TL for dating measurements is that which lies in the narrow temperature window, 290 - 300°C. At this temperature, contamination of the natural TL from signal B is minimised while thermal losses are kept below significant levels.
4.3.2 TL measurement procedures for sediments
The total dose of ionising radiation received by the sediment grains from the time of their deposition to the present day is referred to as the palaeodose. The fine-grain (2-10 µm) fraction of the sediment is preferred for the measurement of the palaeodose because (i) they are most likely to have been well bleached, (ii) their radiation dose rate can be more reliably assessed, and (iii) the TL measurements performed on them are less variable than those of coarser grain sizes. Fine grains are extracted from the sediment and several portions of them are prepared for TL measurement.
Four basic types of procedure are carried out on these portions. Some are left untreated before TL read-out, and these provide the measurement of the natural TL. Others are given artificial radiation doses in the laboratory in order to induce additional amounts of TL, and their TL measurements are referred to as first glows. Others are read out after a week's exposure to daylight, and these yield the bleached, or residual, TL intensity. The last category also receive a prolonged exposure to daylight, but are subsequently given artificial radiation doses in order to regenerate their TL from its residual level; TL measurements on these portions are termed regenerated glows.
Examples of some of these glow curves are shown in figure 2. They comprise natural glow curves, labelled N, measurements of the residual TL after bleaching (R), and regenerated glow curves (R1 - R4) obtained after giving radiation doses of 15, 30, 60 and 120 Grays (Gy), respectively, to bleached portions. (For clarity, examples of first glows have not been drawn.) As explained in the previous section, the TL recorded in the narrow temperature range, 290 - 300°C, is selected for the evaluation of the palaeodose.
Figure 2. Glow curves of TL intensity vs temperature from fine grains of a sediment sample. Shown are (i) the natural TL (N) of the untreated sample; (ii) the residual TL (R) following bleaching of the sample by exposure to daylight; and (iii) regenerated TL (R1 - R4) following irradiation of the bleached sample with increasing doses of ionising radiation.
In figure 3, the selected TL intensities are plotted against artificial radiation dose. The large reduction in TL that results from the bleaching exposure is apparent in the comparison of the natural and residual levels. The data points, R1 - R4, (crosses) correspond to the regenerated TL intensities indicated in figure 2. Also given in figure 3 are the first glow intensities (squares) which show the additional TL induced in the natural sample when it is irradiated.
Figure 3. Growth curves of TL vs radiation dose, illustrating (i) the natural TL intensity of the untreated sample; (ii) the increase of TL intensity when the natural sample is irradiated (first glows); (iii) the level of the residual TL following bleaching of the sample; and (iv) the regeneration of TL (R1 - R4) by irradiation of the bleached sample. The methods for palaeodose evaluation and for determination of the natural regeneration dose (NRD) are also shown.
Two techniques are used for evaluating the palaeodose. The first method is by extrapolation of the first glow measurements to the level of the residual TL. It can be seen from the graph of regenerated TL vs dose in figure 3 that it is not justified to extrapolate the first glow data points with a straight line. The adopted method assumes that the growth of TL in the sediment during burial followed the same form as the regenerated TL growth curve. Firstly, the residual and regenerated TL data points (R and R1 - R4) are fitted with a suitable mathematical form, as shown in figure 3. This growth curve is then manipulated by shifting it horizontally along the dose axis and adjusting its vertical scaling until it gives the best fit to the natural and first glow data points. The palaeodose is obtained where this new curve meets the residual TL level (see figure 3). Note that the adjustment of the vertical scaling is necessary to make allowance for any change in TL sensitivity to radiation produced by the bleaching exposure.
This first method of palaeodose evaluation is entirely similar to that used for burnt flint and calcite TL dating except that for sediments the first glow growth curve is extrapolated to the residual TL level rather than to zero.
The second method is much simpler in conception and gives a more precise evaluation. It consists of calculating the dose where the regenerated growth curve crosses the natural TL level. This dose is termed the natural regeneration dose (NRD). An inspection of figure 3 should make clear that, if the slopes of the regenerated and first glow growth curves are the same, then the palaeodose (determined by the first method) and the NRD will be equal. The two growth curve slopes are the same if the laboratory bleaching causes no change in TL sensitivity. The accumulated evidence of slope comparisons shows that there is no measurable sensitivity change of the 260°C TL signal when it is bleached. The regeneration method is therefore taken as a valid technique for evaluating the palaeodose.
It needs to be emphasised that the procedures described above are specific to the dating signal (i.e. signal A in figure 1). The measured TL signal, however, is invariably a mixture of the dating signal and contaminant TL which derives from the 340°C signal. The regeneration method of palaeodose evaluation is not valid when applied to highly contaminated TL measurements. Levels of the 340°C signal must therefore be kept low for the TL date to be meaningful. In all but a very few sediments, the TL selection procedure described above is able to achieve the necessary degree of signal purity.
4.4. Limitations on the Age Range of Datable Sediments
In general, the maximum ages that can be obtained by TL dating are limited by one of two factors; namely, by saturation of the TL growth, or by instability of the signal. Saturation describes the situation where a sample has received a radiation dose (the saturation dose) beyond which irradiation produces no further increase of the TL signal. For the 260°C signal, the saturation dose is high enough to permit dating throughout the Quaternary. Unfortunately, this potential range of application is serverely restricted because the 260°C signal is unstable.
The instability from which the 260°C signal suffers is of a non-thermal nature. This means that it cannot be overcome by selecting TL of a higher temperature. The instability causes a decay of latent TL intensity characterised by a mean life of 100 ka. Eventually, the rate of TL loss matches the production of TL by radiation, and an equilibrium signal level is reached. The effect on the measured TL dates is to underestimate the true age in a consistent manner. This is shown by the curve in figure 4 which relates the TL date to the true age of the deposition.
Several attempts have been made to overcome this limitation to the applicability of sediment dating. The only alternative to the 260°C signal is the TL which peaks at 340°C. The problem of obtaining a pure measurement of the 340°C signal is similar to that of achieving an uncontaminated sample of the 260°C signal. However, mathematical analysis of the measured glow curves reveals the behaviour of this higher temperature TL. In this way, it is found that the growth of the 340°C signal saturates at a radiation dose of approximately 300 Grays, which is typically the dose received by a sediment of 100 ka age. In practice, it would be difficult to obtain a precise date measurement using this signal where the palaeodose was more than 200 Grays. The early saturation of the 340°C signal, coupled with its difficult bleachability, makes its use as a dating signal very unattractive.
Limitations on the maximum age of datable sediments which are set by signal instability and saturation are absolute, and cannot be surmounted by any valid measurement procedure. In this context, a valid TL date measurement is one in which the observation of the dating signal is substantially unaffected by the presence of the contaminant TL. If this condition is not met apparent TL age values which greatly exceed the limitations can be obtained. Measurements of this type are characterised by their dependence on experimental conditions (e.g. temperature and wavelength selections) which alter the relative proportions of the two signals.
Allowance is made for the instability of the dating signal using the curve shown in figure 4. This curve relates TL dates and true ages, and therefore bears some similarity to the radiocarbon calibration procedure. Figure 4 shows examples of TL dates, drawn next to the vertical axis, being converted to true ages, alongside the horizontal axis. The effect of this conversion on dating precision is described in the following section.
In young sediments, especially those having low dose rates, the natural TL intensity may be similar to the residual level (i.e. to the TL remaining unbleached in the newly deposited sediment). For this reason, precise dating of deposits under 3,000 years old is often difficult. However, significant improvements in the dating of young sediments can be achieved using techniques which detect the residual level and so identify sediments that were insufficiently bleached during transport.
4.5. Precision of TL Dates on Sediments
Error limits attached to measured TL dates result from uncertainties in the palaeodose evaluation and in the dose rate assessment. The size of the palaeodose uncertainty often reflects the level of reproducibility in the TL measurements, which in turn is related to the mineralogical composition of the sediment and the amount of 340°C signal intruding into the selected TL. As a general rule, palaeodose error limits are usually between ±4% and ±8%.
Uncertainties in the assessment of dose rate are estimated from various factors. These include the homogeneity of the sedimentary deposits, the coherence of the dosimetry data and the likely variations of water content throughout the burial period. Typical error limits attached to dose rate measurements are for most cases in the region of ±9%. Measured TL dates therefore have error limits lying mainly in the range from ±9% to ±12% (for a 68% confidence level).
As described above, TL dates have to be converted to true ages by means of the curve given in figure 4. As examples, TL dates of 20±2 ka, 50±5 ka and 80±8 ka are shown yielding true ages of 22.3 +2.5 /-2.4 ka, 69 +11 /-10 ka and 161 +51 /-34 ka, respectively. Two points about the date correction need to be noted. The first is that the symmetrical error limits of the measured TL date are changed to an asymmetric probability distribution, whose asymmetry becomes more pronounced as the TL date increases. The second point is that the uncertainty in the true age inflates rapidly as the TL date approaches its 100 ka limit. This means that, given the typical size of error limits for TL dates, it is unlikely that TL examination will be able to distinguish between a sediment of true age 150 ka and one of 1 million years. Thus, a practical upper limit of 150 ka is set for the true age of sediments that can be usefully dated.
Figure 4. Calibration curve giving the dependence of TL date on true age. Examples of TL dates, next to the vertical axis, of 20±2 ka, 50±5 ka and 80±8 ka are converted to true ages of 22.3 +2.5 /-2.4 ka, 69 +11 /-10 ka, and 161 +51 /-34 ka, respectively, along the horizontal axis.