set_parent("thesis.Rmd")

\chapter{Methodological background, Research design and analytical procedure of the Luminescence dating}

\section{Introduction}

To evaluate my hypothesis and to establish a general chronology of subsistence from 3,400 to 2,000 BP, I used organic geochemistry and luminescence dating methods on the pottery excavated from three major inland sites in the central part of the Korean Peninsula. In the Korean archaeology, the pottery is one of the main objects for the archaeological analysis, being abundant in the Korean Peninsula in almost every archaeological assemblage in sites that post-date 6,000 BP. This abundance has allowed archaeologists to develop a detailed Korean archaeological chronology based on the pottery shape, size and decoration. Though this intensive chronology-building has much contributed to the Korean archaeology, almost no attention has been given to analyzing the fabric of pottery itself. This is a surprising omission and represents a serious gap in our understanding of the prehistoric technology and subsistence. The above methods allow us to identify what was stored and cooked in the pots as well as to date them directly, so that we can understand how subsistence changed over time. Accordingly they let me directly test the hypothesis posited in the previous chapter: that there was utilization of a wider range of resources among ancient farmers in the central part of the Korean Peninsula between 3,400 and 2,000 BP and rice seems to have played no more than a minor role in subsistence during this period. In this chapter, I will discuss the methodological background, research design and analytical procedure of the luminescence dating. I will elucidate some of the main principles of the luminescence dating and its application history to the Korean archaeology. I will also describe the laboratory processes in detail.

\section{Luminescence dating in archaeology}

In terms of the pottery chronology, archaeologists have used stratigraphy that indicates depositional events: when the artifacts were buried together, not specifically when they were manufactured. Dating these depositional events or "occupations" [@Dunnell1971; @Rafferty2008] is a usual goal but it is not quite same as dating manufacturing events. Archaeologists have not always distinguished occupational events and manufacturing events in practice [cf. @Feathers2009]. In addition to stratigraphy, another method employed by archaeologists was the seriation based on the physical characteristics of the potteries. However, this also has an inherent problem, because transmission of the physical characteristics can occur across space [@Dunnell1970; @Feathers2009]. To ascertain that the seriation is mainly entangled in time, it must be restricted to space. The lack of control over the spatial variation means it is difficult to tell whether there are sequential or special differences between each stage of a seriation. However, in real world archaeology, restricting the spatial variation is not always an easy task, especially when the research area is relatively large. The radiocarbon dating somewhat fitted with those traditional approaches, for this well-known absolute dating method mostly does not date the pottery themselves but nearby organic remains (e.g. Charcoal). This means the dating event inevitably has a variable relationship to the target event of pottery manufacture.

Luminescence dating dates the manufacturing event: when the pottery was made. To understand the chronology of subsistence, what archaeologists need to know is the age of the cooking event. Since the cooking event is more likely associated with the manufacturing event than with the depositional event, luminescence dating is probably the most suitable method for creating subsistence chronology. Luminescence dating provides a robust terminus post quem for the cooking event, in a way that is not possible using radiocarbon dating of organic remains.

\section{Luminescence: the principals} Luminescence dating is an absolute dating method that has been used both intensively and extensively in the field of archaeology and Earth sciences. It is based on the emission of light, luminescence, from minerals. In case of pottery, burnt flints, or burnt stones, the dated event is the last heating of the objects. Another common application is dating sediments. In this case, the event being dated is the last exposure of the mineral grains to light. The age range to which the method can be applied is from a century or less to over one hundred thousand years.

Luminescence dating utilizes the radioactive isotopes of elements such as uranium (U), thorium (Th) and potassium (K) [@Feathers2003]. Radioactivity is ubiquitous in the natural environment. Naturally occurring common minerals such as quartz and feldspars act as dosimeters, showing the amount of radiation to which they have been exposed [@Duller2008]. A common characteristic of these naturally occurring minerals is that when they are exposed to the energy emitted by radioactive decay, they tend to store some proportion of it within their crystal structure. The minerals accumulate this energy as their exposure to radioactive decay continues through time. When this energy is released at some later date, it takes the form of light. This light is what we call luminescence.

Luminescence is explained by the solid state energy band theory [@Aitken1985; -@Aitken1998; @McKeever1997]. The interaction between radiation and the crystal structure provides energy to electrons that can be raised from the valence band to the conduction band. Because of this stage, electrons become trapped within the crystal. In the ideal situation, electrons cannot be trapped within the crystal structure, but their trapping is possible because of defects within the structure. Electrons may be stored (and accumulated) at these defects for a certain period. By the time these electrons are released, they lose the energy delivered by the radiation, and may emit a part of that energy in the form of a single photon of light [@Duller2008].

The reason why we can use this phenomenon for dating lies in the fact that this energy stored in minerals can be reset by two processes. The first process is heating the material to the temperature above about 500°C: the process that occurs in a hearth or kiln during firing of pottery. The second is exposure to daylight, as may occur during erosion, transportation, or deposition of sediments. Either of these processes releases any existing energy, and thus set the ‘clock’ to zero [@Duller2008]. Therefore, in the luminescence dating, the event being dated is the last resetting of this clock, either by heat or light.

Measurement of the brightness of the luminescence signal can be used to calculate the total amount of radiation that the sample absorbed during the period of burial. If this is divided by the amount of radiation that the sample receives from its surroundings per year, it will give the duration of time for which the sample has been receiving energy: the age [@Duller2008].

$$\text{age} = \frac{\text{total amount of radiation exposed during burial (equivalent dose)}}{\text{amount of radiation received each year (dose rate)}}$$

There are a number of naturally occurring minerals that emit luminescence signals, including quartz, feldspars, and calcite. Among them, quartz and feldspar are the most suitable and ubiquitous material for dating [cf. @Feathers2003; -@Feathers2009]. The luminescence age is the period of time that has passed since the sample was heated or exposed to daylight. The age is given as the number of years before the date of measurement. Since there is no designate datum for luminescence ages, the date of measurement must be noted. The term BP (before present) should never be used for luminescence ages, for BP designates the specific datum point and is only proper for radiocarbon ages.

The energy that is stored within minerals' crystal structure can be released using a number of laboratory methods.

\subsection{Thermoluminescence}

Heating the sample at a certain rate from the room temperature up to 700$^{\circ}$C releases the trapped electrons within the crystal structure. The resulting signal from this process is called thermoluminescence (hereafter TL). Typically the TL signal comes with a series of peaks (Figure \ref{TL-signal}). Each peak may indicate a single type of trap within the mineral, and commonly the signal comprises several traps. Although it is not always possible to identify the source of electrons precisely, in most cases TL signal observed at the highest temperature originates from the trap that is deepest below the conduction band (more energy is required to release electrons from deeper traps, and therefore this occurs at higher temperature).

A typical thermoluminescence signal (commonly referred to as "glow curve") that shows multiple traps [@Duller2008; cf. @Feathers2003: p. 1495]  \label{TL-signal}

\subsection{Optically stimulated luminescence}

A second way of releasing the electrons stored within minerals is exposing them to the laboratory light [@Huntley1985]. As soon as the mineral is exposed to light, the luminescence is emitted from the its grains. The signal is termed optically stimulated luminescence (hereafter OSL) and Figure \ref{OSL-signal} shows the signal from quartz during the stimulation. As the measurement continues, the electrons in the traps are emptied away and the signal starts to decrease drastically (Figure \ref{OSL-signal}).

A typical optically stimulated luminescence signal from quartz grains [@Duller2008]  \label{OSL-signal}

A similar signal is observed from other minerals including feldspar. However, OSL signal from feldspars decreases more slowly than that from quartz (Duller 2008). Unlike TL, OSL signal does not shows multiple traps. Thus, before measuring the luminescence signal, it is important to thermally pretreat the sample to make sure that the measured signal comes from the deepest traps. This is achieved by heating the sample before measurement so that the shallow traps (whose electrons are unstable over the burial period) are emptied, leaving only the electrons in deeper, stable traps - this heating is called a preheat [@Duller2008: p. 6; @Feathers2003].

The light used to stimulate the minerals is restricted to a certain range of LED lights. The diodes emitting blue light are most widely used type for generating OSL signal from both quartz and feldspar. Another method of stimulation is using the LEDs that emit light beyond the visible part of the light spectrum: infrared stimulated luminescence (hereafter IRSL). IRSL is only observed from feldspars, for quartz does not produce the IRSL signal when the sample is in the room temperature [@Duller2008]. Using these different characteristics of quartz and feldspar, a method for assessing the purity of quartz separated from feldspar for the luminescence measurement can be provided.

\section{Limits of the luminescence dating}

\subsection{Resetting of the signal} The archaeological value of the age obtained from the luminescence dating is determined by whether the resetting event is related to the archaeological event of interest. This means the investigator has to carefully consider the possibility of insufficient exposure or other exposures during the post-depositional process. For heated materials, the most crucial issue to consider is whether the sample was heated to a temperature high enough, and for a period of time long enough, for the trapped electron population to be completely removed. For unheated samples (mostly sediments), the important factors are the time and intensity of the light to which they were exposed during the depositional process. Inadequate exposure to daylight leaves a residual population of trapped electrons.

\subsection{Accuracy and precision}

Limitations on the precision of luminescence ages have been mentioned. When uncertainties in the measurement of the dose rate (often the issues related to the water content) and equivalent dose are combined, errors on luminescence ages normally range from 5 to 10%, including both random and systematic sources of error [@Duller2008]. In archaeological settings, understanding the archaeological context of the site and linking the date with it is a truly important factor for increasing both precision and accuracy. Barnett [-@Barnett2000] used the TL dates of the pottery from later prehistoric Britain to define the typological framework for that period. She found that where a diagnostic form and surface decorations were present, the correlation between the luminescence ages of potteries and the ages from other independent methods was high.

\subsection{Upper and Lower age limits}

The upper and lower age limits to which the luminescence dating is applicable vary from one place to another, and normally depend on the characteristic of the luminescence signal and does rate. The upper age limit is generally governed by the saturation of luminescence signal. At first, the luminescence signal increases almost linearly, but at some point the traps within the crystal structure where electrons can be stored become full. From here, the luminescence signal grows more slowly, until all the traps become full. When the luminescence signal ceases to grow despite continuous exposure, this is what we call saturation. This saturation determines an upper limit of the luminescence dating. Since the point at which saturation can be observed varies from one sample to another, it is impossible to give the precise upper limit to the age that can be obtained. Related to this, Wintle and Murray [-@Wintle2006] suggest that it is reasonable to work in the range where the natural signal is 85% or less of the maximum luminescence signal obtainable. Just as the upper limit, the lower age limit is also difficult to define. The lower limit is mostly controlled by two factors: (1) how well the luminescence signal was reset at the time of the event being dated and (2) the luminescence sensitivity of the mineral being studied.

\section{Luminescence dating and its application to the Korean archaeology}

The luminescence dating is a technique for dating once-heated or -exposed to sunlight materials, and is used by archaeologists primarily to date ancient ceramics and sediments [@Feathers2003]. This technique can measure the time that has elapsed since the last exposure to heat and light of the materials constituting the object. As this exposure event generally occurred when the pottery were made, the luminescence dating is ideal for dating archaeological ceramics [@Feathers2003]. The optically stimulated luminescence dating (hereafter OSL), infrared stimulated luminescence dating (hereafter IRSL), and thermoluminescence dating (hereafter TL) methods employed for dating ceramics have been quite common in Europe and the United States for nearly two decades, but they are yet to be widely used in Korea. Given the abundance of ceramics in Korean archaeological records, it is surprising that the luminescence technique has not been more frequently employed. Though it has been mentioned considerably since its initial introduction [@Choi2006; @Kim2009], it has been used mainly in the field of geology [@Bang2009]. In archaeology, after its applicability was considered [@Hong2001], it has been employed to date several archaeological features including Bronze Age sediments [@Lim2007], Paleolithic sediments [@Kim2010a], historic hydroponic farm [@Hong2003], and potteries from the historic Three kingdom period [@Hong2001a; @Kim2012]. Probably the scarcity of archaeological luminescence dating in Korea may be attributed to the uncritical acceptance of the relative chronologies. I partially agree to the detailed relative chronologies based on the decoration and style of potteries and their serviceable nature [@Lee2008; @Bae2007]. However, since these typological datings tend to ignore spatial variation, their accuracy could therefore be compromised in any particular location. In this regard, the typological dating has its uses, but the verification using luminescence is a prudent approach.

Of course, the primary purpose of the luminescence dating in this research is to investigate the role of the intensive rice farming and to establish the chronology of subsistence strategies over time by correlating the dates it obtained with the results of the organic geochemical analysis. However, with a systematic application of the luminescence dating, I was also able to grasp a glimpse of a more reliable chronology which can be easily applied to other archaeological studies. In 2011, I dated one potsherd from the archaeological deposit in Hongseong city, central part of the Korean Peninsula. Using the thermoluminescence method, I was able to confirm that the potsherd was from the proto-historic period (280 $\pm$ 86AD; U2516 in Table \ref{lumi_eg_date}).

\begin{table}[h] \centering \begin{tabular}{@{}lp{1cm}p{1.5cm}p{1.75cm}lllp{1.75cm}@{}} \toprule Lab. No & Depth (m) & Water Content (\%) & Dose rate (Gy/ka) & TL (De) & OSL (De) & IRSL (De) & Age \ \midrule U3045 & 0.36 & 20.4 & 5.532±0.277 & 8.712±0.91 & 8.586±0.331 & 7.215±0.361 & 280±86 AD \ & & & & 11.665±1.423 & & & \ \bottomrule \end{tabular} \caption{The result of the luminescence dating (The dose rates are rounded to two decimal places, but the calculation of the total dose rate was carried out prior to rounding)} \label{lumi_eg_date} \end{table}

All the samples for my research was dated at the Luminescence Dating Lab, Department of Anthropology, University of Washington, under the direction of Dr Jim Feathers. The luminescence dating method enables the evaluation of the time that has passed since the mineral grains were last exposed to daylight or heated to a few hundred degrees Celsius. Generally, as at the lab of the University of Washington, the method uses an optically and thermally sensitive light or luminescence signal emitted by minerals such as quartz and feldspar. For dating, the amount of absorbed energy (luminescence signal) per mass of mineral (1 J/kg= 1 Gray) due to the natural radiation exposure since the last zeroing - known as the equivalent dose - is determined by comparing the natural luminescence signal of the sample with that which is induced by the artificial irradiation [@Preusser2008]. The time having passed since the last daylight exposure/heating (the date of the sample) is obtained through dividing the palaeodose by the dose rate, the latter representing the amount of energy deposited per mass of mineral by the radiation exposure on the sample over a certain time [@Preusser2008]. The potsherds in this thesis were dated by using this formula, and all the three methods, TL, OSL, and IRSL were applied. For a further clarification, the dates from the luminescence dating were correlated with those from AMS radiocarbon dating.

\section{Analytical procedure}

The luminescence dating method was developed in an archaeological context, in Europe in the 1960s and 1970s, as a method of dating heated materials, primarily ancient ceramics and potteries [@Feathers2003]. It has been applied to a wide range of Quaternary researches such as those on landscape evolution, palaeoclimate, archaeology, and has been being refined since its early days. It dates the past exposure to heat and light, and because the events of this exposure are the actual events archaeologists are interested in, it has a strong merit over other dating methods [@Feathers2003]. In other words, in the luminescence method, the dating event is often the target event that archaeologists are looking for. In this thesis, the luminescence dating was applied to seven archaeological ceramic samples.

\subsection{Sample preparation - grain size}

For the luminescence dating, determining the grain size is quite important, for it occasions diverse advantages/disadvantages as well as different methods. Generally, fine grains (1-8 um) are more abundant than coarse ones; and they can be analyzed with samples of relatively small amount. They also require a relatively simple sample preparation process, and rely less on the external dose rate, which is often problematic in a complex ceramic environment. However, if samples include feldspar grains (which cannot be separated from other grains during the sample preparation procedure), one has to deal with the high fading rate of feldspar [@Wintle1973].

One of the biggest advantages of using coarse grains (180-212 um) is the single grain analysis, which can be done only with coarse grains. Quartz grains are generally used for the analysis of coarse grains, because of their well-known properties and low fading rate. Since it is possible to minimize feldspar inclusion during the sample preparation process of coarse grains, we do not have to consider the fading of feldspar as a major variable. Also, because of the larger grain size and etching process during the sample preparation, the contribution of alpha radiation (which has a short range: 50um) is minimal. This is a huge merit, for alpha radiation is much less effective in producing luminescence than beta and gamma radiations. In case of analyzing fine grains, this 'low alpha efficiency' must be considered. However, using coarse grains for the analysis requires a complicated sample preparation process and a larger amount of samples. Also, it cannot be totally exempted from the high fading rate, because feldspar has to be used for the single grain analysis in some cases (feldspar typically has a bright luminescence signal, which enables dating older deposits than with quartz) where quartz shows an extremely low luminescence signal [@Preusser2008]. It has also been verified that the quartz of volcanic origin may show anomalous fading, just like feldspar [@Bonde2001; @Tsukamoto2007]. In this thesis, fine grains were used for the analyses, because of their small sample size and advantages that I have mentioned above.

\subsection{Glassware and reagents}

All glassware was washed with Decon 90 (Decon laboratories), rinsed four times in distilled water. Analytical grade reagents (typically $\geq$ 98% purity) were used throughout.

\subsection{Dose rate measurement}

The dose rate is the amount of energy deposited per mass of the mineral by the radiation exposure of the sample over a certain time [@Preusser2008]. For the dose rate measurement, the exposed parts of the potsherds were used (0.5-1 g). The dose rates were determined by alpha counting (Low level alpha counter 7286: Little more Science Engineering Co., DayBreak alpha counter 583: DayBreak), beta counting (Beta multi counter system RIS\O\ GM-25-5: Ris\o\ National Laboratory), and flame photometry (Flame Photometer PFP-7: Jenway).

The water absorption percentages of the samples were measured. This is quite important for calculating the dose rate, as the attenuation of radiation is much greater if the sample is filled with water [@Preusser2008]. For measuring the water absorption percentage, the sample was saturated with deionizing water for several days. Then, its surface wetness was removed by gently dabbing it with a wet paper towel; and then it was immediately placed on the scale to weigh it. After the sherd was dried in a 50$^{\circ}$C oven for several days to record its weight in its dry state. The water absorption percent is calculated as W = [(S/D)/D]*100, where S is the saturated weight and D, the dry weight.

Some component of the dose rate is produced by the ionizing cosmic radiation, and could be different by the geographic location and burial depth of the sampled material (Prescott and Hutton, 1994). All information related to the latter points was obtained from the excavation records of the sites where the samples came from. Alpha counting gives the current alpha activity rate. And based on this rate and the assumption of secular equilibrium, one can calculate the beta and gamma dose rate. However, by using the beta counter and flame photometry as well, we can enhance the validity of the total dose rate measurement (flame photometry is used to measure K content and the beta counter is used to assess the accuracy of alpha counting and flame photometry measurements). This sort of advantage is available only if we utilize multiple tools at the same time.

\subsection{Equivalent dose measurements}

For measuring the equivalent dose (paleodose) of the pottery samples, TL (Thermo luminescence; DayBreak 11000 Automated TL system), OSL (Optically stimulated luminescence; RIS\O\ TL/OSL system DA-15), and IRSL (Infrared stimulated luminescence; RIS\O\ TL/OSL system DA-15) were utilized. Artificial laboratory irradiations were given by the Irradiator type 721/A (Little more Science Engineering Co.) and RIS\O\ TL/OSL system DA-15. For beta radiation, Sr-90/Y-90 beta source, calibrated against a Cs-137 gamma source, was used. Am-241 source was used for Alpha irradiation. Fine grains (1-8 um fractions) were used for dating. The grains were obtained from the core part of the potsherds more than 2 mm away from any exposed surface. This was done by drilling, using tungsten carbide drill bits.

For the TL analysis, the equivalent dose was determined by the slide method to obtain both of the advantages of the additive dose method and the regeneration method [@Aitken1985; @Prescott1993]. The slide method can deal with the matter of extrapolation as well as the change in sensitivity simultaneously. These two problems cannot be solved at the same time in case of using either the additive dose method, or the regeneration method solely. The regeneration curve can be used to define the extrapolated area and can be corrected for sensitivity change by comparing it with the additive dose curve. The equivalent dose is taken as the horizontal distance between the two curves after a scale adjustment for sensitivity change.

OSL and IRSL on fine-grain (1-8µm) pottery samples are carried out on a single aliquot following procedures adapted from Banerjee et al. [-@Banerjee2001] and Roberts and Wintle [-@Roberts2001]. The equivalent dose is determined by the single-aliquot regenerative dose (SAR) method [@Murray2000]. The SAR method measures the natural signal and the signal from a series of regeneration doses on a single aliquot. The method uses a small test dose to monitor and correct for sensitivity changes brought about by preheating, irradiation or light stimulation. SAR consists of the following steps: (1) preheat, (2) measurement of the natural signal (OSL or IRSL), (3) test dose, (4) cut heat, (5) measurement of test dose signal, (6) regeneration dose, (7) preheat, (8) measurement of the signal from regeneration, (9) test dose, (10) cut heat, (11) measurement of the test dose signal, (12) repeat of the steps from 6 to 11 for various regeneration doses. Usually a zero regeneration dose and a repeated regeneration dose are employed to insure the procedure is working properly. For fine-grained ceramics, a preheat of 240$^{\circ}$C for 10s, a test dose of 3.1 Gy, and a cut heat of 200$^{\circ}$C are currently being used, although these parameters may be modified from sample to sample.

For OSL and IRSL, the luminescence was measured on a Ris\O\ TL-DA-15 automated reader by a succession of two stimulations: first 100 s at 60$^{\circ}$C of IRSL (880nm diodes), and then 100s at 125$^{\circ}$C of OSL (470nm diodes). Detection is effected through 7.5mm of Hoya U340 (ultra-violet) filters. The two stimulations are used to construct IRSL and OSL growth curves, so that two estimations of equivalent dose are available. Feldspar usually involves anomalous fading and only feldspar is sensitive to IRSL stimulation. The rationale for the IRSL stimulation is to remove most of the feldspar signal, so that the subsequent OSL (post IR blue) signal is free from anomalous fading [@Roberts2001]. However, feldspar is also sensitive to blue light (470nm), and it is possible that IRSL does not remove all the feldspar signal. Some preliminary tests in our laboratory suggested that the OSL signal does not suffer from fading, but this may be sample specific. The procedure is still undergoing study.

As I mentioned above, for dating fine-grained samples, one has to deal with the low alpha efficiency. This is taken into account by determining the alpha efficiency factor: "b-value" (Huntley et al. 1988). It has been known that the alpha efficiency varies between quartz and feldspar (Huntley et al. 1988). The typical b-value of quartz and feldspar is respectively about 0.5 and more than 1.5. For TL, the alpha efficiency is determined by comparing additive dose curves using alpha and beta irradiations. The slide program is also used in this regard, taking the scale factor (which is the ratio of the two slopes) as b-value (Aitken 1985). The results from several samples from different geographic locations show that OSL b-value is less variable and centers around 0.5. IRSL b-value is more variable and is higher than that for OSL. TL b-value tends to fall between the OSL and IRSL values. Currently, measuring the b-value for IRSL and OSL is in process by giving an alpha dose to aliquots whose luminescence have been drained by exposure to light. An equivalent dose is determined by SAR using beta irradiation, and the beta/alpha equivalent dose ratio is taken as b-value. A high OSL b-value is indicative that feldspar might be contributing to the signal and thus subject to anomalous fading.

\subsection{Determining the age}

The time having passed since the last daylight exposure/heating of the pottery sample (Hereafter: age) was calculated through dividing the palaeodose by the dose rate. The final date of the sample was obtained through calculating the average of the three dates from TL, OSL, and IRSL. Normally, when conducting the luminescence dating on a pottery sample, its associated sediment is required for the precise dose rate measurement. However, since there was no associated sediments on my samples, I relied on an average of sediment dose rates determined in other parts of Korea [@Hong2003; -@Hong2001a; @Kim2010a; @Kim2012; @Lim2007]. The age and error for both OSL and TL are calculated by a laboratory constructed spreadsheet, based on Aitken [-@Aitken1985]. All error terms are reported at 1-sigma.

\section{Summary}

In this chapter, I have discussed the methodological background, research design and analytical procedure of the luminescence dating. Some of the main principles of the luminescence dating and its application history to the Korean archaeology were elucidated. I also described the laboratory analytical process in detail.



SeungkiKwak/Kwak_S_PhD_thesis documentation built on May 9, 2019, 1:22 p.m.