These notes are intended as a guide for archaeologists and geologists to the method of thermoluminescence (TL) dating. The date measurement combines TL examination of the dated sample with information about the site conditions surrounding the sample. Even in cases where the site need not be visited in the course of the dating project, the participation of the date user is still an important requirement. These guidance notes provide information that will help to promote the success of the collaboration.
The notes begin with a description of the phenomenon known as thermoluminescence. This is followed by an explanation of how it is applied to date measurement and a list of the datable materials. In the second section, the conditions required for date measurement, and the factors limiting it, are listed. The relevant considerations are the isolation of the dating signal, its initialisation in antiquity, the limitations due to TL saturation and stability, and the need to assess the radiation dose-rate. Some of the research methods which have been employed to investigate the potential of the various applications are described in the next section. These include the plateau method for detecting composite TL emissions, techniques for separating the TL signals, investigations into the initial state of the TL signal, observations of TL saturation, and tests of the signal stability. In the final section, some elements of the date measurement are explained, including how the palaeodose is evaluated, the use of the plateau test, and what tests need to be made on the reliability of the date measurements.
1.1. Principles of TL Dating
1.1.1 What is TL?
All buried materials are exposed to a constant flux of ionising radiation. This radiation originates from naturally occurring radioactivity, which is present in all deposits, and from cosmic radiation. When crystalline materials are exposed to ionising radiation, a redistribution of electrical charge takes place within the crystal. Much of the displaced charge finds its way back to its original state within a short space of time, but a small fraction of it can become trapped for long periods in higher energy states. In this way the crystal can be said to retain a memory of the ionising radiation to which it has been exposed.
The extra energy that the crystal contains as a result of the radiation exposure can be released by heating the material. When heat is applied, some of the released energy appears in the form of light, causing the material to luminesce. If the crystal is then cooled and re-heated it does not re-emit light, because the energy excess which produced the first emission has now been released from the crystal. This effect is known as thermoluminescence (TL). It appears only while irradiated crystals undergo a progressive temperature increase, and should not be confused with incandescence, which is the light radiated continuously by hot bodies.
The graph which traces the variation of the TL intensity with increasing temperature is known as the glow curve of the sample. In general, the glow curve exhibits one or more peaks, which occur whenever the increasing thermal energy of the crystal becomes sufficient to release electrical charge from the various traps in which it is held. Figure 1 shows a set of glow curves exhibiting the sharp peak at 275°C which is characteristic of stalagmitic calcite. Several portions of the same sample were measured to obtain this series of glow curves. Some of the portions were given artificial radiation doses before their measurement, while others were left unirradiated. The glow curves illustrate how the intensity of the TL peak grows as the radiation exposure increases. The TL emission produced by the unirradiated portions is known as the natural TL, since it results from the radiation dose accumulated by the sample in the natural environment over geological time. TL measurements such as those shown in figure 1 provide a means of determining the amount of this radiation dose, which is termed the palaeodose.
Figure 1. Characteristic glow curves of stalagmitic calcite, showing the growth in the intensity of TL emission produced by exposing the material to increasing amounts of ionising radiation. Measurements such as these enable the radiation dose received by the sample over archaeological or geological time scales to be determined.
1.1.2 Basis of the TL date measurement
As shown above, the amount of light emitted during TL depends upon the total radiation dose to which the crystalline material has been exposed. If the material has undergone an event in the past which has caused a release of the excess energy which gives rise to TL, the natural TL measured at a later time provides a measure the palaeodose received in the intervening period. The amount of the accumulated palaeodose is proportional to both the rate of radiation absorption by the material, and the time that has elapsed between the initialising event and the TL read-out. The following simple equation relates these quantities:
1.1.3 Datable materials
The three main categories of materials which may be dated by TL are flint (or stone), stalagmitic calcite, and sediments. Specific details of the dating of these materials are given in Chapter 2 (Heated Flint and Stone), Chapter 3 (Stalagmitic Calcite), and Chapter 4 (Sediments).
Ceramics, while they are also datable by TL, are not often useful in archaeological applications, since their typology is usually a more precise indicator of age. However, a very similar application is the dating of burnt daub, where the age might not be otherwise known.
1.2 Necessary Conditions and Limiting Factors
A number of conditions must be satisfied before TL can be applied as a dating technique. These requirements, and the factors which limit the age range of the TL method, are set out in the following sections.
1.2.1 Isolation of the TL dating signal
In general, the observed TL glow curve is the sum of several luminescence producing processes. Each crystalline phase may produce one or more TL signals. In addition, it is impossible to entirely eliminate certain mineralogical impurities which, although they may be minor constituents of the TL sample, can contribute disproportionately to the TL glow curve. Therefore, methods for separating and isolating the desired dating signal are usually needed.
As illustrated above in figure 1, each TL signal is characterised by the temperature at which its glow curve peaks. In addition, the wavelength (or colour) of the light is also characteristic of the signal. Thus, each signal displays a distinctive graph of TL intensity versus wavelength, which is referred to as its emission spectrum. In most TL applications, the desired signal can be adequately isolated by selecting the appropriate wavelength and temperature of the emitted light. The ability to separate signals by temperature selection is a unique and valuable advantage of the TL method.
Methods for analysing glow curves and separating the TL signals contained in them are described below.
1.2.2 Initialisation of the TL intensity
For TL dating to be possible, the crystalline material must have undergone an event which erased its memory of previous radiation exposures. In the case of flint, this erasure occurs if it is heated, either by man or in a natural fire. Stalagmitic calcite can also be dated because, when its crystalline structure is newly formed, it contains no significant quantity of latent TL. For a body of sediment, the initialising event is the exposure to daylight which often occurs at the time of deposition, since light is almost as efficient as heat in releasing trapped charge. In all cases, it is the date of the initialising event which is measured by the TL technique.
1.2.3 Saturation of the TL intensity
The graph which records the growth of TL intensity versus increasing radiation dose is referred to as its growth curve. The quantity of TL emitted by a crystalline material cannot increase indefinitely with additional radiation doses. It eventually reaches a limit where further irradiation produces no increase of TL intensity and the growth curve becomes flat. Then the TL is said to be saturated, and the dose at which this occurs is called the saturation dose. This behaviour is illustrated in figure 2.
Figure 2. The growth curve of a sample of flint, showing the declining rate of growth of the TL intensity as the sample is irradiated with increasingly large doses. The practical dose limit for palaeodose evaluations using this material is indicated where the TL intensity is approximately 50% of the saturation level.
Clearly, palaeodoses can only be measured if they are comfortably below the saturation dose. This effect therefore imposes an upper age limit upon TL applications. Saturation doses vary greatly between different materials and samples, and it is often impossible to predict when this limitation will operate. However, as a general rule, it must be expected that materials subjected to high rates of radiation exposure will meet saturation earlier than those in low dose-rate environments.
1.2.4 Stability of the TL signal
As another necessary condition, it is required that the excess energy retained by the crystal does not leak away to any significant degree during the interval of time to be measured. If leakage does occur, the measured TL age will be less than the true age.
The plateau test is one means of detecting unstable TL, but is only reliable where the cause of the instability is thermal de-trapping. There exist other mechanisms that can cause the latent TL intensity to decay. Present knowledge does not allow the extent of the losses to be calculated from first principles. Ultimately, a TL signal can only be shown to be stable if it yields correct dates for known age samples.
1.2.5 Assessment of the radiation dose-rate
Equally as important as the preceding conditions is the requirement that the rate of radiation dose to which the material has been exposed can be assessed to the desired accuracy. Ionising radiation exists in the form of alpha, beta and gamma rays, and originates from radioactive decays of the naturally occurring elements, uranium, thorium and potassium. Cosmic rays also contribute to the total level of radiation.
Alpha, beta and gamma rays have very different ranges in geological materials. Alpha rays have the shortest range (approximately 0.03 mm); beta rays traverse up to 3 mm in solid matter; while gamma rays penetrate about 30 cm. If one considers a flint flake, of a few centimeters size, which is buried under more than 0.3 m of soil, then almost all the alpha and beta radiation that it receives originates from within the flint itself, while nearly all the gamma dose is derived from the surrounding soil. The assessment of the dose-rate to the sample from gamma and cosmic sources is referred to as the environmental dosimetry. By removing the outer 2-3 mm of the flint, one is left with a portion whose alpha and beta dose derives solely from the radioactivity contained in the sample itself. The assessment of this dose is termed the internal dosimetry.
For many dating applications, the environmental dose contributes most of the total dose received by the TL sample, and emphasis is therefore placed on its accurate measurement. The preferred method of assessing the environmental dose-rate is by direct measurement of gamma and cosmic radiation levels at the burial site of the TL sample. A detailed account of on-site dose-rate measurement is given in Chapter 5 (Environmental Dosimetry).
1.3. Research Methods in TL Dating
As a first stage in exploring the potential of any TL dating application, research needs to be directed to answering two fundamental questions. These are: "How many TL signals are present?", and "What are their properties?". In the following sections, some of the research methods that have been used to answer these questions are explained.
1.3.1 Plateau method for detecting composite TL emissions
Composite TL emissions are those which receive contributions from two or more luminescence producing processes. In some cases, it is not obvious from the appearance of the TL glow curves whether or not they are composite. If, in any TL dating application, the signals composing the luminescence have significantly different properties, there exists the potential for major error in the evaluation of the palaeodose. It is therefore important to employ an effective method for detecting composite TL emissions. The plateau principle provides a means of achieving this.
When measuring a TL glow curve the dating specialist is free to set the conditions of a number of experimental variables. Most importantly, the specialist may vary (i) the procedure for preparing the sample for TL examination, (ii) the range of wavelengths detected by the TL apparatus, and (iii) the temperature interval selected for the palaeodose evaluation. The plateau principle states that, in the vicinity of the conditions selected for the palaeodose evaluation, the final result should not depend on the precise settings of the variables. (Clearly, if by making small adjustments to the measurement procedure significant changes are produced in the apparent age of the sample, the measurement cannot be considered to be reliable.)
Historically, the first application of the plateau principle was to identifying the part of the natural glow curve which had remained stable during the burial of the sample. The use of the "plateau test" is described below. However, even after selection of the thermally stable TL, it is often found that the apparent value of the palaeodose depends on the temperature or the wavelength of the selected TL, or the method of sample preparation. The varying palaeodose is a reliable indicator that the TL emissions are composed of more than one signal.
In order to understand the reason for this, it must be appreciated that the experimental conditions control the ratios in which the signals appear in the glow curves. Thus, varying the selected temperature range alters the ratio between signals that display different glow curve shapes. Similarly, varying the selected wavelength band changes the ratio between signals with different emission spectra, and alterations to the method of sample preparation can discriminate between signals originating in different mineral phases. All such variations therefore affect the signal composition of the measured TL glow curves and the temperature regions selected from them. (For examples, see figures 3 and 4 below.)
When applied in a systematic way, changes in signal composition can be used to discover the number of TL signals composing the emissions and the properties that the signals possess. The method is described in the following section. Here, it is sufficient to point out that, where two signals differ in their properties, the evaluated palaeodose will depend on the ratio in which the signals are present. In this context, the critical properties are the form of the growth curve, and the change of growth curve slope occurring during the palaeodose evaluation procedure. Differences in these properties result in a failure of the plateau principle, which in turn reveals the composite nature of the TL emissions.
The case of TL emissions from sediments supplies the best example of the plateau principle in action. Sediment TL is composed largely of two signals but, because the glow curves of these signals overlap to a high degree, the shape of the composite glow curve does not always reveal its true structure. It is found that, for relatively young sediments (less than 50 ka) the measured palaeodose is not sensitive to the experimental conditions. However, for older sediments, different measurement conditions (such as are likely to exist between different laboratories) produce varying magnitudes for the apparent age of the material. This is easily explained as follows. One of the signals possesses a low saturation dose which means that, in sediments older than 50 ka, the growth rate of its TL intensity with radiation dose is declining to zero. This signal also differs in the way its growth curve changes on bleaching. The value of the measured palaeodose therefore varies widely depending on whether the non-saturated signal or the saturated signal dominates in the composite growth curve. The manner in which the palaeodose varies with temperature, wavelength and sample preparation (and additionally on bleach treatment) is in accordance with the known properties of the two signals, and demonstrates clearly the composite nature of sediment TL.
1.3.2 Techniques for separating TL signals
The plateau principle is useful for detecting when the observed TL emissions possess a complex structure, but gives few clues about the number of TL signals that are present or what properties they possess. This section describes techniques useful for separating the signals and examining them individually.
In general, the most effective way to separate the components of a TL glow curve is to exploit the differences between their temperature and wavelength characteristics. The standard TL equipment allows the emitted light to be passed through a filter, so that one particular band of wavelengths may be preferentially selected. Much research has been performed simply by observing the alteration of glow curve shape which follows a substitution of this filter. With more specialised equipment, it is possible to measure TL output as a function of both temperature and wavelength, and thereby obtain the emission spectrum of the material simultaneously with its glow curve.
In simple glow curves, such as those shown above in figure 1, the dating signal is readily recognised by the shape and position of the peak in TL intensity. In the measurement of this glow curve, the height of the peak has been enhanced relative to the background of unwanted emissions by passing the light through a coloured filter which transmits wavelengths characteristic of calcite TL while absorbing other signals. A further improvement in the purity of the calcite signal can then be obtained by selecting only the TL which falls within a narrow temperature range around the position of its peak.
Unfortunately, not all TL glow curves are as simple as those of calcite. The case of sediment TL illustrates the difficulties that can be encountered. While the TL of sediment is not essentially more complex than that of other materials, the degree to which the components of the TL overlap in temperature and wavelength presents a more problematical situation.
Figure 3. Temperature and wavelength characteristics of the two principal TL signals emitted by sediments. The centre panel shows the emission spectra (TL intensity versus wavelength) of Signals A and B. The right-hand panel displays the TL glow curve obtained when the blue region of the spectrum is selected. In comparison, the TL glow curve in the left-hand panel illustrates the increased ratio of Signal A to Signal B when ultra-violet emissions are selected.
Sediment light emissions are dominated by two overlapping TL signals, referred to as Signals A and B. For dating purposes, the properties of Signal A are far more useful than those of Signal B. In figure 3, the central panel contrasts the emission spectra of the two components. The degree of overlap between these spectra limits the ability of wavelength filters to separate the signals. However, it can be seen that the optimum filter for enhancing Signal A is one which transmits ultra-violet emissions and absorbs blue light. The left-hand panel of figure 3 shows the TL glow curve (A+B) which is obtained when such a filter is used. For comparison, the right-hand panel shows the typical appearance of a sediment TL glow curve when it is observed through a blue-transmitting filter.
Clearly, in order to maximise the intensity of Signal A relative to that of B, the selected TL should be both at ultra-violet wavelengths and at low temperatures. Unfortunately, the TL emitted at the lowest temperatures is not stable enough to be used for date measurement. For this reason, the optimum selection is confined to a narrow temperature range close to the 270°C to 300°C region. Under these experimental conditions an adequate degree of purity can be achieved for the measurement of Signal A.
While temperature and wavelength selections alone can achieve a sufficiently pure sample of the dating signal for most purposes, it is sometimes desirable to effect a complete separation of the glow curve components. This is especially useful in the context of researching the detailed properties of individual signals. The shapes of the glow curves of signals A and B, shown in figure 3, were derived by a systematic study which exploited not only the differing temperature and wavelength characteristics of the signals but also their reactions to different methods of sample preparation.
Figure 4. Variations of the natural glow curve shape obtained from one sample of sediment by altering the conditions of sample preparation and wavelength selection. The measured glow curves are labelled A+B. By a procedure of subtracting one glow curve from another the shapes of the two signals which compose them have been deduced. These shapes are labelled A and B.
Figure 4 shows the variations in the natural glow curve of one sediment that were achieved using different sample preparations and wavelength selections. The two glow curves on the right were measured after the fine-grain sample had been treated with acid, while the pair on the left are from the untreated sample. Within each pair, alterations of glow curve shape have been brought about by the use of different wavelength filters. It is clear that all the measured glow curves can be constructed from combinations of just two basic shapes, demonstrating that only two TL signals (A and B) are required to account for them. The two shapes have been derived mathematically by subtracting one glow curve from another.
By a simple extension of the subtraction method, it is also possible to measure the properties of each individual signal when acted upon by such agents as light exposure, radiation dose and pre-heat treatment. These analyses thus provide all the information that is required by the specialist to select the signal that has optimum properties for dating, and to devise the most suitable technique for palaeodose evaluation.
1.3.3 Initial state of the TL signal
Assuming that a given TL signal can be observed in a sufficiently pure condition, the next most important requirement of the dating technique is that the initial TL intensity is known. The response of the signal to the initialising event depends largely on the agent which is responsible. For heated materials, such as flint and stone, the initial state of the TL signal is in no doubt, because heating completely erases the record of all previous radiation exposure. When dating stalagmitic calcite, which has not been heated, it is necessary to rely on the experience of research, which shows that newly formed travertine produces a negligible amount of TL when it is measured.
In the case of sediment dating, in which the initialising event is an exposure to light, there is always some uncertainty about the degree of exposure that occurred at deposition. Research has shown that TL intensity is progressively reduced by increasing exposures, but that it is never completely removed. Therefore, the level of residual TL which survived the deposition event cannot be accurately known, and this creates a corresponding uncertainty in the date measurement. In comparison to Signal B, Signal A is more rapidly reduced by light, and to a much lower residual level, and this is the main reason why Signal A is chosen as the dating signal. Additional dating strategies include a preference for fine-grained deposits which are more likely to have experienced long transport times.
1.3.4 Observation of TL saturation
The question of the saturation of TL signals is not always easily answered by research. In simple glow curves, in which the TL signals are well separated in temperature, the growth of each peak as the radiation dose to the sample is increased can be clearly discerned. The growth curve of increasing TL intensity versus radiation dose eventually bends into a horizontal line, showing that the material has reached its saturation dose (see figure 2).
However, in composite glow curves, where two or more TL signals overlap, the growth curves are similarly complex. In this case, saturation of the TL intensity does not occur until all components of the growth curve have saturated. Composite TL structures are generally recognisable by the manner in which the shape of the glow curve undergoes major alteration as radiation dose is increased. (Note, however, that some simple TL signals display moderate shifts in peak temperature on irradiation.)
Sediment TL exemplifies the problems of observing TL saturation in composite glow curves. Because a wide section of the temperature scale is occupied by two overlapping signals, the growth curves of the individual signals are difficult to observe directly. Instead, a mathematical analysis of the composite TL glow curves is necessary in order to separate the two signals and discern their properties (as described above). From such analyses, it is found that, while Signal A has a very high saturation dose, Signal B saturates at a much lower dose. Because growth curve measurements form the basis of all palaeodose evaluations, it is important that the TL structure which determines their behaviour is well understood.
1.3.5 Tests of TL stability
There is more than one decay mechanism which may cause the loss of latent TL intensity during the burial time of the sample. One such process is thermal de-trapping, in which thermal energy is the agent which releases the displaced electrical charge from the traps in which it is held. (This is the same process that operates when TL is observed.) The effects of thermal instability are usually easy to recognise by the routine application of the plateau test.
The rate of de-trapping is determined by the quantity of available thermal energy, which is in turn dependent upon the ambient temperature of the sample during its burial. By holding the material at a constant elevated temperature, a measurable loss of the TL intensity may be observed over a reasonably short interval of time. If this experiment is repeated at a number of different temperatures, the data may be extrapolated to obtain the rate of de-trapping at the temperatures which the sample experienced during its burial. Hence, an estimate can be obtained of the degree of TL loss during the burial period.
It should be emphasised that such experiments only test for TL instabilities that are due to thermal de-trapping. The presence of other decay processes can be revealed through the dating of known age samples.
1.4. Date Measurement Procedures
The previous section has described the principal methods for separating the available TL signals, and for investigating some of their properties. The information gained by these researches permits the specialist (i) to decide which TL signal is most useful for the purposes of date measurement, (ii) to deduce the optimum experimental conditions for observing that signal, and (iii) to devise a plausible procedure for using the signal to evaluate the palaeodose. Elements of the dating process are described in the following sections.
1.4.1 Evaluation of the palaeodose
Essentially, the palaeodose is evaluated by comparing the natural TL intensity of the sample with the increase of TL output induced by known amounts of additional radiation. By extrapolating the growth curve until it intersects the initial TL intensity of the sample, the dose accumulated since the initialising event can be found.
In this procedure a complication arises because, in general, the growth curves are not straight lines. While it is possible to observe the form of the growth curve for TL intensities greater than the natural level, the manner in which the TL grew from its initial state is not revealed. This difficulty is answered by measuring a second growth curve, using portions of the sample in which the initial state of the TL has been recreated by heating or exposure to light. The form of this second growth curve is then taken as the correct line to use when extrapolating the first growth curve. It should be noted that this assumption is ultimately untestable by scientific investigation, but can only be justified through the dating of known age samples.
The methods of palaeodose evaluation appropriate to different materials are described more fully in Chapter 2 (Heated Flint and Stone), Chapter 3 (Stalagmitic Calcite), and Chapter 4 (Sediments).
1.4.2 Plateau test for thermal stability
An important difference exists between the natural glow curve and the TL induced by a laboratory irradiation. Because of thermal de-trapping, the TL which is emitted at low temperatures (typically below 200°C) is unstable. In natural samples this TL has decayed away during burial, and is therefore absent from the glow curve. In contrast, artificially dosed samples emit this low temperature TL because the interval between irradiation and measurement is too short for decay to have a significant effect. This difference forms the basis of the plateau test, which identifies the temperature region of the natural glow curve in which the TL intensity is unaffected by thermal decay. The name of the test derives from the constant height of the graph of palaeodose values versus temperature in the region of the stable TL, as illustrated in figure 5. The ability to recognise thermal instability in this way is a unique and valuable advantage of the TL method.
Figure 5. Variation of the palaeodose as evaluated at different temperature co-ordinates of the TL glow curve. The zero starting value of the palaeodose reflects the complete absence of natural TL at low temperatures due to its decay by thermal de-trapping. The palaeodose plateau at high glow curve temperatures indicates the region of stable TL.
It should be noted that the interpretation of the plateau test is straightforward if only one TL signal is present in the glow curve, or, when more are present, if all the signals possess similar growth curves. In more complex situations, as found for instance in old sediments, variations of the palaeodose with temperature and other experimental variables require a more detailed analysis.
1.4.3 Tests of the reliability of date measurements
In the final stage of the dating procedure, the evaluation of the palaeodose is combined with an assessment of the rate at which the sample accumulated that dose, to determine the time which has elapsed since the initialising event. It has been emphasised above that experimental research is unable to establish a priori the validity of all aspects of the palaeodose evaluation. In addition, there are assumptions made in the dose-rate assessment that are not open to investigation. For both these reasons, the TL technique, while it has a sound theoretical basis, may only be ultimately justified by testing it on known age samples, and through comparisons with other dating methods.
By means of such tests, the reliability of TL for dating flint and calcite is strongly supported for the period back to Oxygen Isotope Stage 7, and less firmly established for approximately twice that age. Similar tests of the TL method of sediment dating show that, due to a non-thermal instability of the dating signal, it is limited to a maximum age of 150 ka. Further details on the age ranges of the various TL applications are given in Chapter 2 (Heated Flint and Stone), Chapter 3 (Stalagmitic Calcite), and Chapter 4 (Sediments).