2.1.1 Petrographic Analysis

Petrographic analysis, also known as thin-section analysis or optical microscopy, is a conventional, yet powerful, analytical technique extensively employed in the scientific investigation of archaeological ceramics. The analysis is carried out on a thin section, which is examined under a special microscope, the so-called polarized light microscope. A thin section is a thin, flat slice with typically 0.03 mm thick, cut from a ceramic sample and mounted onto a glass slide using a specific epoxy resin. The principles of petrographic analysis are based on the identification of mineral inclusions through their known optical properties. Under the microscope, the so-called rock-forming minerals, which are usually the main components of the inclusions, display specific optical properties such as birefringence, pleochroism, twinning, relief, extinction, and other optical features, each unique to a particular mineral type (Figure 1a-d). These diagnostic features facilitate the precise identification of the individual minerals. Similarly, based on these principles, rock fragments that may be included in ceramics can also be identified.

Figure 1 Examples of thin-section photomicrographs of different types of ceramic materials: (a) Late Bronze Age coarse ware from Central Anatolia. (b) Late Chalcolithic coarse ware from Seville, Spain. (c) Early Bronze Age fine ware from Northern Mesopotamia. (d) Early Bronze Age fine ware from Southeastern Anatolia. (e) SEM image of Late Bronze Age ceramic from Plain Cilicia, Southern Anatolia.

Figure 1Long description

Images (a-d) show ceramic samples viewed under a polarising microscope, highlighting the variations in the composition of the fabric. Samples (a) and (b) are photomicrographs of coarse-textured ceramics containing abundant, clearly visible mineral and rock fragments. These inclusions can be precisely identified petrographically and provide valuable information for the determination of the provenance and insight into the manufacturing techniques. Samples (c) and (d) show fine-textured, homogeneous ceramics with minimal or no visible inclusions. Petrographic analysis of such fine ceramics is of limited interpretative value because of the absence or scarcity of diagnostic mineral grains and rock fragments. Figure (e) shows an SEM image of a ceramic sample. Textural features such as vitreous, partially fused or fused textures (degrees of vitrification) can provide valuable insights into ceramic production processes and are particularly useful for estimating firing temperatures.

The analysis involves a qualitative examination of the composition of mineral and rock fragments, as well as the micro-textural features of ceramic fabric. This is a rapid, cost-effective, and widely available method, serving as a valuable tool for the archaeometric investigation of ancient ceramic materials. It is particularly capable of examining a wide range of material properties, especially in coarse-grained ceramics. Although this method can be applied to explore various archaeological questions, a common focus in many studies using petrography is determining the geographic origin of the production place of the vessels. This is the only method that enables the identification of rock inclusions in ceramics. Through this method, it is possible to precisely assign the raw material source used in ceramics to a specific area or geological unit, especially in cases where specific rock types are included in the ceramics (Kibaroğlu & Thumm-Doğrayan Reference Kibaroğlu and Thumm-Doğrayan2013).

Another main aspect of petrographic analysis is the reconstruction of production processes. The microscopic observation of microtextural features, such as grain size, quantity, shape, and distribution, along with other features like grain orientation, is important for gaining information about the techniques used in production. This may include the material properties of the clay, the strategy for resource choice, and the specific processing techniques applied to the clay paste, such as tempering or purification of raw clay. Petrographic observations can also provide insight into the shaping techniques of vessels (Thér Reference Thér2016). Furthermore, matrix features such as the optical behaviour of the clay matrix (whether it is optically active or inactive), the degree of vitrification, and the size and morphology of voids are also useful for estimating firing conditions (Braekmans & Degryse Reference Braekmans, Degryse and Hunt2017). By combining petrographic examinations with other mineralogical techniques like Scanning Electron Microscopy (SEM) or X-ray Diffraction (XRD), more accurate information regarding production technology can be obtained.

The insights gained from such analysis can significantly contribute to our understanding of past societies by providing information on the production places of questioned ceramic groups, manufacturing techniques, technological advancements, trade patterns, cultural interactions, and socioeconomic dynamics. Its cost-effectiveness and widespread availability have made it a widely applied technique in this field. However, there are some methodological limitations, especially when applied to fine ceramics: its efficiency may be limited in identifying provenance and in studying production technology. However, petrographic analysis of fine ceramics may still provide some valuable information; for example, distinguishing between calcareous and non-calcareous clays. Additionally, it enables a broad estimation of the firing temperature based on the observation of the vitrification grade of the ceramic matrix, thus providing clues about the potential firing temperatures of the sample. This method is also useful for studying slips and glazes as it allows us to observe their internal microstructure and examine the interaction between the glaze and the clay body. It also enables us to identify unfused inclusions and new reaction phases, as well as determine whether pigments are applied over or under the glaze (see Section 6).

Petrographic analysis is inherently destructive; it involves cutting a sufficiently small piece from the sample to create thin sections. In such cases, its application to very small samples or to precious artefacts from museums is limited or not feasible. One further drawback of this method is that it requires specialized knowledge in geology, particularly in the fields of petrology and sedimentology, and hence, requires intensive training.

2.1.2 Scanning Electron Microscopy (SEM) and SEM-EDX

The scanning electron microscope (SEM) is a powerful imaging technique that is widely used for obtaining high-resolution images and detailed information about the surface and near-surface structures of a broad spectrum of solid materials. SEM utilizes a focused beam of electrons to scan the surface of a specimen. This electron beam interacts with the atoms on the surface of the sample, generating signals that can be detected and transformed into high-resolution images (Ul-Hamid Reference Ul-Hamid2018). This is a useful technique applied to studying a wide range of archaeological materials, involving ceramic and inorganic painting decoration of ceramics (Tite Reference Tite1992; Burnstock & Jones Reference Burnstock, Jones, Creagh and Bradley2000; Froh Reference Froh2004).

Measurement in an SEM is inherently non-destructive and is generally operated in a vacuum (typically 10–5 mbar). However, as the sample chamber has a limited volume, only small objects, such as arrowheads, coins, or small shards, can primarily be analysed without sample preparation. Alternatively, in a specially designed large sample chamber, non-invasive measurement can also be achieved (Barnes Reference Barnes1991). In the case of analysing a ceramic body (paste), the sample should be prepared, thus making the measurement destructive in this instance. Generally, a polished surface of ceramic is prepared and coated with a thin layer of gold or carbon (typical thickness of 50 nm) prior to the measurements, to prevent the build-up of electric charge on the specimen, when bombarded with electrons (Burnstock & Jones Reference Burnstock, Jones, Creagh and Bradley2000).

There is a wide range of potential applications for analytical SEM in the field of archaeological materials. SEM provides insights into the physical properties and microstructural characterization of both raw materials and finished ceramics. It also contributes to provenance studies; however, its primary use in ceramic studies is the investigation of production technologies, specifically the estimation of firing temperature (Tite Reference Tite1992). This is achieved through the characterization of the various stages of vitrification, which depend on the firing temperature. The high-resolution SEM images (Figure 1e), typically on the order of 1–2 nanometers, enable the examination of the microstructure of ceramics, allowing for the estimation of the potential or maximum firing temperature of the ceramic (Tite Reference Tite1992; Froh Reference Froh2004). Modern SEM is commonly combined with energy-dispersive X-ray analysis (EDX), also known as SEM-EDX. This combination allows for the determination of the elemental composition of selected spots on samples, enabling the identification of specific inclusions in the matrix or areas on the surface, such as different pigments. SEM-EDX also facilitates the identification of elemental variation in a specific surface area on the sample, known as elemental mapping (Froh Reference Froh2004; Aprile et al. Reference Aprile, Castellano and Eramo2019).

Due to its flexibility, especially when combined with EDX (Energy Dispersive X-ray Spectroscopy), this technique is highly versatile. It can operate both as a destructive and non-destructive method, depending on the research question and/or the area to be analysed. This capability makes SEM-EDX a preferred technique for comprehensive material characterization in ceramic archaeometry. SEM analyses are also essential for analysing slips and glazes to observe the microstructure and analyse the slips/glaze themselves (see Section 6).

2.2.1 X-ray Diffraction

X-ray diffraction (XRD) is a well-established and widely used analytical method for determining the mineralogical composition of various inorganic materials, including archaeological ceramics. This method works by irradiating the sample surface with incident X-rays and measuring the intensities and angles of the X-ray reflections, and through Bragg’s Law, which explains the relationship between the angles of incidence and the resulting diffraction patterns (for technical details, e.g. Heimann Reference Heimann and Hunt2017).

XRD analysis on powdered samples (Powder XRD) requires sample preparation and is therefore destructive. A small portion of the sample from a ceramic shard, about 0.2–0.5 g, needs to be fine powdered, usually in an agate mortar, until the particle size is uniform and less than 20 µm. The mineral composition of the powdered sample is then analysed. Based on the generated signals, the mineral phases are identified using the International Centre for Diffraction Data (ICDD) PDF database.

XRD is also widely used to investigate the production processes of ceramics, with particular emphasis on the raw materials used and the firing temperature and atmosphere (Maggetti Reference Maggetti, Olin and Franklin1982; Cultrone et al. Reference Cultrone, Rodriguez-Navarro, Sebastian, Cazalla and De La Torre2001; Trindade et al. Reference Trindade, Dias, Coroado and Rocha2009). The method also is useful to examine the possible post-depositional alteration processes of ceramics (Heimann & Maggetti Reference Heimann and Maggetti1981; Schwedt et al. Reference Schwedt, Mommsen, Zacharias and Buxeda i Garrigós2006). Another area of its application is the analysis of mineral-based painting decoration of ceramics (Moon et al. Reference Moon, Kim, Nam and Cho2021).

One of the strengths of this method in ceramics study is that it enables the identification of fine crystalline phases. This is particularly useful for fine ceramics, where petrographic methods provide limited information (see Section 2.1). It allows for the investigation of the mineralogical composition of the clay. In low-fired ceramics, it may also be possible to determine the initial clay minerals of raw clay. XRD is a useful tool, especially in the estimation of the firing temperature of ceramic vessels which is one of the key questions largely subjected in many archaeometric studies, as this provides information to understand the technological procedures applied by the ancient potters. This concept relies on the presence or absence of certain minerals, referred to as indicator minerals, of sample under investigation. Such indicator minerals are for example calcite, hematite, mullite, gehlenite, or diopside. During the firing process, specific chemical reactions occur within the clay paste. Depending on factors such as firing temperature, firing atmosphere, mineralogical composition, and the fineness of the raw clay, some minerals disappear at specific higher temperatures, while others are newly formed. By identifying such indicator minerals, it is possible to determine the firing temperature (or temperature interval) of the ceramic (Gliozzo Reference Gliozzo2020). Besides its primary application in exploring the production processes of ceramics, XRD analysis, especially of fine ceramics, can also provide valuable information for the interpretation of chemical data. This is particularly relevant in regard to provenance identification.

Due to its easy accessibility in many laboratories, cost-effectiveness, and other advantages, powder XRD is one of the preferred analytical techniques applied in ceramic archaeometric studies. XRD is also used to identify crystalline compounds in glazes. It can be performed directly on the surface of the glaze, provided the surface is sufficiently flat. The depth to which X-rays penetrate the glaze varies depending on its composition, typically ranging from a few micrometres to tens of micrometres. This variability can sometimes limit the identification of compounds near the glaze-ceramic interface. Transmission μ-XRD using synchrotron light offers the advantage of accessing different areas across the thin (50–100 μm) cross section of the glaze with micrometre-level resolution (Pradell & Molera Reference Pradell and Molera2020).

2.2.2 Raman Spectroscopy

Among the various analytical techniques used in archaeometric studies for examining mineralogical compositions, Raman spectroscopy has become a significant tool during the last decades for exploring and characterizing a wide range of archaeological materials (Edwards et al. Reference Edwards, Vandenabeele and Colomban2023). Raman spectroscopy identifies molecular structures and composition of sample in different states – solids, liquids, and gases – making it a highly versatile instrument. The technique uses monochromatic light, typically a laser as the light source, which interacts with the molecules in the sample, a small part of the light scattered due to the excitation of the molecular vibrations. This scattered light, known as the Raman scattering effect, is characteristic for specific molecules present and their vibrational modes. By analysing the Raman scattering patterns, a fingerprint of the sample’s composition can be determined, hence enables the identification of various constituents, the study of their structures, and the quantification of their abundance (e.g., Edwards et al. Reference Edwards, Vandenabeele and Colomban2023 for details).

The technique is inherently non-destructive or minimally invasive, as it allows samples to be analysed in their natural state without the need for extensive preparation such as cutting, powdering, or dissolving. However, in the case of laboratory-based analysis, if a sample is too large to fit into the sample compartment, a small portion of sample (c. 1 g) may need to be removed from the larger sample. A notable advantage of Raman spectroscopy, compared to other mineralogical analytical techniques such as X-ray diffraction, is its ability to identify both crystalline and amorphous components, as well as organic-based materials. Furthermore, owing to continuous technological advancements, Raman spectroscopy has further advantages in terms of smaller instrument size and improved transportability, facilitated by better laser sources, electronics, and CCD detectors (Vandenabeele et al. Reference Vandenabeele, Edwards and Moens2007). Its flexibility in application and the analytical performance have contributed to its growing popularity, leading to its widespread application across a vast range of research areas, including archaeometry, art history, conservation and restoration, and many other fields. In archaeometric studies, Raman spectroscopy has been extensively applied to analyse a wide variety of material categories, such as metals, vitreous materials like glass and glazes, pigments, ceramics, biomaterials, and others, to address a different set of research questions (e.g., Edwards et al. Reference Edwards, Vandenabeele and Colomban2023).

Concerning the archaeometric analysis of ceramic materials, its previous application has focused more on surface decorations, such as painting pigments and glazes (see Section 6). Especially in the study of paintings, Raman spectroscopy offers some benefits compared to other methods, such as XRD. This technique is able to identify not only mineral-based pigments (inorganic crystalline phases) but also organic-based pigments. This can be particularly important in cases where there is no prior information about the possible pigment types (organic or inorganic origin). Though the application of Raman spectroscopy to the analysis of the ceramic body composition, for example to identify mineral inclusions in ceramics, has been used (e.g., Wopenka et al. Reference Wopenka, Popelka, Pasteris and Rotroff2002; Medeghini et al. Reference Medeghini, Mignardi and Vito2013), its use for the analysis of the ceramic body is still limited, compared to the study of surface decorations. As this method is capable of identifying the mineralogical phases of ceramics, for example, on a small powder samples or other sampling forms, it can provide valuable information, particularly on the production techniques. This may include identifying mineral phase which were newly formed during the firing process, such as mullite, diopside, wollastonite, or other minerals, as well as amorphous components that may be critical to determining the firing setting of ancient ceramic production.

Identifying the material composition in a sample using Raman spectroscopy is typically based on the comparison of the characteristic vibrational spectra of the sample to reference spectra in a database. For this purpose, there exist freely accessible databases containing reference spectra for a wide range of materials, including minerals, ceramics, pigments, glasses, organic compounds, and other sample types. The most frequently used databases are the RRUFF database (http://rruff.info/) and the IRUG (Infrared and Raman Users Group) database (http://www.irug.org/). There are also several other open-access libraries available (for further details and references, see Edwards et al. Reference Edwards, Vandenabeele and Colomban2023: 37–38). When integrated with other commonly used analytical techniques – such as X-ray fluorescence (XRF) and scanning electron microscopy coupled with energy-dispersive X-ray spectroscopy (SEM-EDS) – Raman spectroscopy can provide additional advantages for comprehensive materials characterization, thereby enabling the exploration of a broader spectrum of archaeological research questions.

Raman spectroscopy is today considered as one of the most suitable instrumental techniques. This is mainly due to the sensitivity, high analytical performance, and non-destructive analysis capabilities of the technique, as well as the flexibility of its application for the analysis of diverse material groups. It is to be expected that the continuing development of new, non-invasive instrumentation and the availability of portable and handheld device setups (see Section 2.4.2) will further contribute to its widespread use in archaeometric studies of ceramic materials.

2.3.1 X-Ray Fluorescence Spectrometry

X-ray Spectrometry (XRF) is a well-established analytical technique widely applied in scientific and industrial fields to determine the major, minor, and trace elements composition of various materials, ranging, for example, metals, glasses, ceramics, rocks, pigments, as well as environmental and biological materials and more. The measurement is based on the secondary X-rays emitted by the sample and thereby the detection of characteristic X-rays by a spectrometer that is used to identify the elements present in the sample and calculate their concentrations. There are several spectrometer techniques in various configurations that were developed and applied for various purposes (Beckhoff et al. Reference Beckhoff, Kanngießer, Langhoff, Wedell and Wolff2006; Pollard et al. Reference Pollard, Batt, Stern and Young2007). Depending on the method of the detection system, this technique is usually divided into two main types: energy-dispersive (ED) and wavelength-dispersive (WD) X-ray fluorescence analysis. EDXRF measures the energy and intensity of the secondary X-rays and is more suited for elements with an atomic number (Z) between 20 and 41. WDXRF measures the wavelengths of the secondary X-rays, and can also measure lighter elements from about Z = 11 (e.g., Schramm Reference Schramm2012; Hall Reference Hall and Hunt2017).

Principally, due to its non-invasive nature, EDXRF is widely used, in particular, different configurations of EDXRF such as Micro-XRF and portable XRF (see Section 2.4.4); both are non-invasive techniques, are widely applied in the analysis of art, museum, and archaeological objects such as ceramic, metal, glass, manuscripts, paintings, icons, and other (Mantler & Schreiner Reference Mantler and Schreiner2000; Donais & George Reference Donais and George2018). Both EDXRF and WD-XRF are two analytical techniques that are able to generate high-quality compositional data and provide a low-cost and rapid measurement of the chemical composition of ceramic materials. Although the samples can be analysed by both methods without sample preparation, for example, in the case of small samples that fit into the sample holder, labour-based analysis can especially benefit from sample preparation for high-quality results. This may include simple cleaning and polishing of the sample surface, powdering, and pelletizing with or without a binder (e.g., ceramic materials), fusing the sample with appropriate flux (ceramics, rocks, ores, etc.).

Both laboratory-based ED XRF and WDXRF methods have been successfully used in ceramic research as shown in numerous studies (e.g., Papachristodoulou et al. Reference Papachristodoulou, Oikonomou, Ioannides and Gravani2006; Kibaroğlu et al. Reference Kibaroğlu, Satır and Kastl2009; Reference Kibaroğlu, Sagona and Satır2011). The measured elements show a high precision, largely even comparable with NAA data (Müller et al. Reference Müller, Hein, Georgakopoulou, Kilikoglou and Kiriatzi2018). Nevertheless, both spectroscopic methods have advantages and disadvantages when compared. For example, EDXRF is considered a fast method that allows the determination of several elements simultaneously with high accuracy. In addition, EDXRF is also cheaper and more easily accessible. However, WDXRF has its own advantages over EDXRF. Elements with lower Z, such as Na, Mg, P, Al, and Si, are best measured using WDXRF spectrometry (Hall Reference Hall and Hunt2017). WDXRF spectrometers can also more readily measure some of the REEs (rare earth elements) that may be significant in provenance identification. WDXRF spectrometers generally have superior detection limits over EDXRF spectrometers (Jenkins Reference Jenkins1999; Janssen Reference Janssen, Janssens and Van Grieken2004).

Both spectroscopic techniques are powerful tools for the examination of ceramic materials. As in the case of other analytical techniques, method selection usually depends on several criteria such as research questions, expected results (quantitative or semi-quantitative analysis, determination of element suites), performance, sample size, cost issues, availability of the technique, and others (Hall Reference Hall and Hunt2017).

2.3.2 Neutron Activation Analysis

Neutron Activation Analysis (NAA) is a well-established and frequently employed analytical technique for determining the elemental composition of various materials, including archaeological ceramic materials (Perlman & Asaro Reference Perlman and Asaro1969; Harbottle Reference Harbottle and Newton1976; Glascock Reference Glascock and Neff1992). This is a nuclear analytical technique that uses the emission of characteristic gamma rays from a sample that has been irradiated with neutrons to determine the concentration of elements in the sample (Neff Reference Neff, Ciliberto and Spoto2000; Glascock & Neff Reference Glascock and Neff2003).

For the measurement of elemental composition, relatively little effort is generally required for sample preparation. The sample can be removed by breaking off a small piece of the sherd and then grinding it, or by scraping, drilling, or burring it from the material. The prepared sample, commonly in the form of powder, is placed on a gamma ray detector and then irradiated. The required sample size depends on the homogeneity of the sherd. For inhomogeneous samples, for example, coarse ceramics, more material may be needed. Although NAA can also measure very small samples (30 mg), there may be some loss of precision in the case of such small sizes. Therefore, for an adequate measurement, a sample size of between 100 and 150 mg is typically needed (Neff Reference Neff, Ciliberto and Spoto2000; Sterba Reference Sterba2018).

NAA has a number of advantages over most other analytical methods, especially in regard to archaeological materials such as ceramic provenance studies. NAA is capable of measurements of a large range of major, minor, and trace elements with high precision, accuracy, and sensitivity, which may be important for reliable provenance identification. This is a significant advantage for archaeological studies, as it allows the analysis of small size or valuable artefacts, without leaving remarkable traces. A further advantage is that measurement is nearly free of any matrix interference effects. The comparability of NAA data from different laboratories is possible without major conversions, so existing databases can be used as a reference group for provenance studies (Neff Reference Neff, Ciliberto and Spoto2000). This is especially beneficial for ceramic research, where there is a large NAA database available (ceraDATFootnote ¹, MURRFootnote ²).

NAA is a cost-intensive technique, though its expenses can vary depending on several factors, including reactor costs, equipment expenses, reference materials, the number of samples, and overall labour costs. One of the main disadvantages of NAA is that it can only be performed in a research reactor. Due to the limited availability of such reactors, only a few research groups at well-funded government laboratories have been able to explore new applications for nuclear techniques. As a result, there are only a limited number of laboratories around the world that perform NAA; notable examples are the Missouri University Research Reactor (MURR) and TU Wien Atominstitut, Austria. This has led to a kind of monopoly, with research dependencies on these few laboratories. A significant problem is also nuclear waste disposal. Though there are relatively small amounts of radioactive waste produced, its disposal is challenging. As Neff (Reference Neff, Ciliberto and Spoto2000) forecasted 20 years ago, as a result of environmental concerns related to nuclear waste and the construction of new reactors, a declining trend in the use of NAA can be observed, and this method is increasingly being replaced by other methods, such as Laser Ablation Inductively Coupled Plasma Mass Spectrometry (ICP-MS).

2.3.3 Inductively Coupled Plasma Mass Spectrometry

ICP-MS (inductively coupled plasma mass spectrometry) is a powerful analytical technique, widely employed in diverse research fields including geoscience, archaeological science, and industrial sectors. Due to its ability to perform rapid multi-element determinations at the ultra-trace level and technical improvements in instrumentation, ICP-MS has been one of the fastest-growing analytical techniques in recent years. The basic principle of measurement is to ionize atoms using suitable ionization methods, separate the ions by their mass-to-charge ratio (m/z), and then record the mass spectrum using a recording device (e.g., Pollard et al. Reference Pollard, Batt, Stern and Young2007; Thomas Reference Thomas2013).

Although the first ICP-MS was introduced in the early 1980s, and was subsequently applied in various fields, such as in geology, its application in ceramic archaeometry has become increasingly popular, especially in the recent decade (e.g., Giussani et al. Reference Giussani, Monticelli and Rampazzi2009; Resano et al. Reference Resano, Garcia-Riuz and Vanhaecke2010; Golitko & Dussubieux Reference Golitko, Dussubieux and Hunt2017; Kibaroğlu et al. Reference Kibaroğlu, Kozal, Klügel, Hartmann and Monien2019; Williams et al. Reference Williams, Sharratt and Banikazemi2023). There are two main types of ICP-MS instruments mostly used in ceramic archaeometry: Digestion (or Solution) ICP-MS and Laser Ablation-ICP-MS. Both methods have their own applications and advantages in the context of ceramic analysis.

The Digestion-ICP-MS requires the sample to be in a liquid form; therefore, for measurement of solid samples, for example, ceramic, a small part of the sample, typically 100 mg is dissolved in a strong acid, such as nitric acid or hydrochloric acid, generally using microwave technique to obtain an accurate bulk chemical composition of the material (Kennett et al. Reference Kennett, Sakai, Neff, Gossett and Larson2002). Although the Digestion ICP-MS is a highly sensitive method and needs a small sample size (~100 mg) for measuring the major, minor, and all geochemically relevant trace elements (>40), which offers an advantage in the ceramic provenance study compared to other methods, for example, NAA, however, its wider application in ceramic archaeometry is limited. This is especially due to the health risks associated with acid treatment of the samples and waste disposal, also a time-consuming sample preparation (Little et al. Reference Little, Kosakowsky, Speakman, Glascock and Lohse2004).

Laser Ablation (LA) ICP-MS: Combined with a laser-based sampling system, called Laser Ablation (LA), ICP-MS has become increasingly the most extensively used high-resolution analytical technique for multi-element characterization for a wide variety of archaeological materials such as vitreous materials, ceramics, pigments, metals, human remains, and other archaeomaterials (Giussani et al. Reference Giussani, Monticelli and Rampazzi2009; Resano et al. Reference Resano, Garcia-Riuz and Vanhaecke2010; Dussubieux et al. Reference Dussubieux, Golitko and Gratuze2016). The use of LA-ICP-MS in ceramic archaeometry has increased in recent years rapidly, due to technological development and the resulting improvement in analytical performance (Speakman & Neff Reference Speakman and Neff2002; Golitko & Dussubieux Reference Golitko, Dussubieux and Hunt2017; Kibaroğlu et al. Reference Kibaroğlu, Kozal, Klügel, Hartmann and Monien2019; Williams et al. Reference Williams, Sharratt and Banikazemi2023).

The ablated sample from the ablation cell is first transported into the plasma torch by argon or helium gas. The produced aerosol is then ionized in the plasma, passing through the mass spectrometer where the ions are measured by a detector (Giussani et al. Reference Giussani, Monticelli and Rampazzi2009; Thomas Reference Thomas2013 for further details). LA-ICP-MS has a number of advantages over other analytical techniques that make it the method of choice for various fields of archaeological science. Some important advantages of this method, among others within the ceramic examination, are its capability to identify a large number of elements (>40), including major, minor, and trace elements (including Rare Earth Element-REEs), with high analytical performance, which are important criteria for a reliable provenance identification. Major elements can also be identified without external measurement, using the normalization technique (Gratuze et al. Reference Gratuze, Blet-Lemarquand and Barrandon2001).

The surface damage caused by the laser ablation is minimal, in most cases invisible to the naked eye, having a typical diameter of 50–200 µm. Small sherd samples, fitting into the ablation cell, can also be analysed directly, on a cleaned specific surface (typically clay matrix) without sample preparation (e.g., Eckert & James Reference Eckert and James2011; Sharratt et al. Reference Sharratt, Golitko and Williams2015; Shoval Reference Shoval2017). In this way, however, the measured element composition is a ‘pseudo-bulk’ composition. To address this, and to obtain bulk chemical data that is representative of the sample, the measurement needs to be performed on press-pellet (6–8 mm in diameter) that is perpetrated from 0.5 to 1.0 g homogenized ultra-fine sample powders (Kibaroğlu et al. Reference Kibaroğlu, Kozal, Klügel, Hartmann and Monien2019). As the LA-ICP-MS has micrometre spatial resolution (50–100 µm), it is also possible to analyse particular mineral inclusions (Gehres & Querré Reference Gehres and Querré2018) or components included in a ceramic sample. A notable drawback is up to today the limitation of matrix-matched Standard Reference Materials (RMS) for quantitative analyses of archaeological ceramics.

The increasing availability of LA-ICP-MS in numerous laboratories, combined with technological advancements, makes this method appealing for various research areas. The portable laser ablation techniques for sampling (pLA-ICP-MS), a new development (see Section 2.4.5), provide new opportunities in the archaeometry research area.

2.4.1 Portable X-ray Diffraction Analysis (p-XRD)

Portable mobile XRD and laboratory XRD are both used to determine the crystal structure of materials (see Section 2.2.1). However, while the latter offers greater precision and detailed analysis, the former has the advantage of portability and rapid on-site analysis, making it a valuable tool for field work and in situ characterization of materials. However, measurements can take up to 60 minutes due to the limited size of the portable instruments and the resulting need to adjust the placement of components in a slightly less-optimal way. Also the detection limit is quite low, allowing mainly the major phases to be investigated. p-XRD instruments are available from a number of manufacturers and differ in aspects such as measurement mode, calibration, data processing, and overall performance. Additionally, measurement spot size, arrangement of components, and the mounting of the device itself can be adjusted. It is therefore important to be well informed about which application is best suited to the user’s purpose, which will require a certain investment of time in research and training for the user to achieve reliable and useful results. Laboratory XRD typically requires extensive sample preparation including grinding and homogenization to ensure uniformity and accurate results. In contrast, portable XRD requires less sample preparation for polycrystalline materials such as ceramics, often limited to cleaning the surface to remove contaminants. This makes it suitable for objects that are too valuable to allow even a small area of the surface to be removed. However, this means that any surface contamination will inevitably affect the analysis results. In terms of handling, the sample must be placed in a fixed distance in the centre of the goniometric circle, the latter being the part of the apparatus that controls the exact angular position of the sample relative to the beam and the detector. Generally speaking, only one measurement per sample is taken (Pappalardo et al. Reference Pappalardo, Pappalardo and Amorini2008; Romano et al. Reference Romano, Pappalardo, Masini, Pappalardo and Rizzo2011; Nakai & Abe Reference Nakai and Abe2012).

2.4.2 Portable Raman Spectroscopy (p-RS)

Raman spectroscopy (see Section 2.2.2) is used to identify mineral phases in archaeological pottery, with the ability to analyse amorphous materials such as carbon. While laboratory RS instruments are typically large and stationary, requiring a controlled laboratory environment, portable RS instruments are compact, mobile or handheld, allowing easy transport and operation in different locations, including the field or remote environments. The latter are well suited for rapid on-site analysis without the need for complex sample preparation procedures, as surface cleaning is sufficient, while the former offer higher precision and detailed analysis. While mobile instruments can be used flexibly because the probe can be positioned independently of the rest of the necessary equipment, handheld instruments are more suitable for field work. The technology uses a laser to induce Raman scattering, the spectrum of which can be used to identify the molecule in the sample by comparing the visible spectrum with a reference database. A measurement can take between 5 and 30 minutes, depending on the instrument and setup. Some studies used very short measurement times of only 20 seconds with three measurements per sample, holding the instrument at a distance corresponding to the focal point of the instrument. It is worth mentioning that the method is not fully suitable for quantitative analysis and the spectra can be difficult to interpret. In addition, mobile instruments need to be mounted in a stable position, with the ability to move around the sample. As with all methods, it is important to be aware of the specifics of the instrument – in this case, excitation wavelength, fibre availability, and so on. It is also very important to get familiar with the standards of the method and to follow a strict protocol for sampling and data processing. Even if the method is normally non-destructive, it may be minimally invasive if the laser power is too high for the sample. Therefore, sample properties and laser energy must be carefully monitored to avoid destroying parts of the sample (Nakai & Abe Reference Nakai and Abe2012; Vandenabeele & Donais Reference Vandenabeele and Donais2016; Barone et al. Reference Barone, Di Bella and Mastelloni2017; Jehlička & Culka Reference Jehlička and Culka2022).

2.4.3 Portable Fourier Transform Infrared Spectroscopy (p-FTIR)

Portable Fourier Transform Infrared Spectroscopy is used mainly for the study of painted pottery, aiding in the identification of ceramic bodies, glazes, pigments, and firing temperatures. To identify the mineralogical composition not only of crystalline but also of pseudo-amorphous thermal phases, infrared spectroscopy uses a broad spectrum of infrared light to excite the molecules in a sample that absorb the infrared radiation. They begin to vibrate in characteristic ways which are processed, and the interference patterns are analysed and converted into a spectrum by a mathematical transformation (= Fourier transformation). This then provides information about the functional groups and chemical components in the sample, as the visible peaks correspond to the specific molecular vibrations. The mid-infrared range is often applied for excitation, but if molecules with functional groups such as N-H, O-H, or C-H are to be identified, the near-infrared range should also be used. As this method is non-destructive, it is also susceptible to surface contamination and curvature. The measurement time can vary (4 to 8 minutes) with a measurement spot size of approximately 0.6 cm and a collection of at least one spectrum per sample. Again, an in-depth knowledge of measurement parameters and data processing is required to obtain reliable results. Spectra can be improved by using powder samples. This reduces external reflection and improves the signal-to-noise ratio. Knowledge of data processing and interpretation of peaks for identification of composition phases by using literature or reference materials is essential to achieve high-quality research results (Bruni et al. Reference Bruni, Longoni, De Filippi, Calore and Bagnasco Gianni2023).

2.4.4 Portable X-ray Fluorescence Analysis (p-XRF)

Like their laboratory counterparts, mobile and handheld X-ray fluorescence analysers (Figure 2) can be used to determine the chemical composition of pottery samples in the field. This allows the identification of elements present in the pottery, providing information on raw materials and possibly their origin. Therefore, p-XRF can help answer key research questions such as the provenance of pottery, trade and exchange networks, technological advancements, and cultural interactions. However, due to its limited range of detectable elements but short measurement time, this method is often used as a preliminary screening tool or complementary method. While laboratory (see Section 2.3.1) and mobile XRF analysis provide high accuracy and detailed elemental information, they also require samples to be transported to a laboratory involving more extensive sample preparation. On the other hand, handheld p-XRF analysis offers rapid on-site analysis with minimal sample preparation preserving the integrity of valuable artefacts and providing immediate data that can inform ongoing archaeological investigations. Measurement time can – due to application and device – vary between about 6 minutes and 10 seconds. These instruments are available from a variety of manufacturers, each providing their own measurement settings (methods) and data processing algorithms. It is therefore necessary to understand the specific instrument in sufficient detail to be able to select the appropriate measurement parameters and apply the best method of data quality control and processing. While understanding these details can be tricky, the practical operation of the instruments is quite simple: equipped with a simple trigger, the nose of the instrument must be as close to the sample as possible. The surface to be analysed should be as smooth and free of contamination as possible. For this reason, fresh brakes are often produced to allow at least three non-overlapping measurements per sample. However, as with p-XRD, handheld p-XRF can be used in a completely non-destructive manner with the same implications for data interpretation (Potts Reference Potts, Potts and West2008; Frahm & Doonan Reference Frahm and Doonan2013; Shugar & Mass Reference Shugar and Mass2014; Vandenabeele & Donais Reference Vandenabeele and Donais2016).

Figure 2 A variety of producers offer handheld p-XRF devices,

© M. Schauer

2.4.5 Portable Laser Ablation Inductively Coupled Plasma Mass Spectrometry

The portable laser ablation ICP-MS technique is a combination of two different instrumentations that work in different settings. The main difference in this method compared to the standard LA-ICP-MS (Section 2.3.3) is the sampling technique. While in a lab-based setup, a stationary ablation device with a sample chamber of limited size is used, pLA-ICP-MS uses a specifically designed portable laser ablation sampling device, which allows in situ sampling of arbitrarily sized objects in the field and museum setting (Glaus et al. Reference Glaus, Koch and Günther2012; Knaf et al. Reference Knaf, Koornneef and Davies2017). The prime component is the portable LA sampling device, which basically consists of a pulsed laser, an optical fibre attached to a hand-held LA module, a sampling filter mounting, and a membrane pump. For sampling, the LA module is applied to the sample surface, and the produced particles, as aerosol particles, are collected on membrane filters with sub-micrometre pore size. After the on-site sampling procedure, the filters are re-ablated using a standard laser device used in the laboratory with standard settings, the sample is introduced to the mass spectrometer, and the detected element concentrations are measured. Alternatively, from the filter-sampled material, the sample solution is prepared, and subsequent elemental analyses performed with standard ICP-MS (Numrich et al. Reference Numrich, Schwall and Lockhoff2023). An experimental study has presented that quantification of major, minor, and trace elements by means of pLA-ICP-MS is of similar performance as a standard LA-ICP-MS setup. Compared to other portable techniques, such as X-ray fluorescence spectroscopy (XRF), multiple orders of magnitude lower limits of detection can be obtained (Glaus et al. Reference Glaus, Koch and Günther2012). The pLA-ICP-MS technique has already been successfully applied to different archaeological materials, including metals (copper, silver, gold, e.g. Burger et al. Reference Burger, Glaus and Hubert2017; Seman et al. Reference Seman, Dussubieux, Cloquet and Pryce2020), jade, and porcelain (Käser Reference Käser2015). Experimental analysis by Glaus et al. (Reference Glaus, Koch and Günther2012) also demonstrated the suitability of this method for archaeological ceramic materials.

The pLA-ICP-MS opens a novel possibility in the archaeometric analysis of cultural-historical objects, which for various reasons cannot be analysed destructively. Regarding ceramic analysis, its application is still in the initial phase, but it also offers opportunities to analyse the ceramic body itself and also surface decorations (painting and glaze) with high sensitivity. A drawback of this method is its limited availability, as currently there are only four sampling devices available (Knaf et al. Reference Knaf, Koornneef and Davies2017). However, it can be presumed that with the further development of this technique, both internally and in its device components, there will be an increasing popularity.

3.2.1 Raw Material Selection

The first step in the chaîne opératoire is the selection of appropriate clayey raw materials (Gualteri Reference Gualteri2020). These are commonly collected from raw material deposits in the vicinity of the pottery workshop (Arnold Reference Arnold1985). The collected clayey raw materials can be generally categorized as fine-grained sedimentary rock, in which the actual clay fraction commonly constitutes only one part. The clay fraction is defined by grain size (< 2 μm) and prevalent argillaceous particles with platy crystal structure. Apart from these so-called phyllosilicates, which are predominant in the clay fraction, the raw materials contain typically non-plastic accessory minerals with larger grain sizes. Furthermore, it can contain associated non-crystalline phases, such as organic matter (Bergaya & Lagaly Reference Bergaya, Lagaly, Bergaya, Theng and Lagaly2006).

When sufficiently high temperatures are reached during firing, the phyllosilicates decompose and new high-temperature mineral phases are formed. The initial clay paste composition, particularly in terms of type and amount of clay minerals, affect the mineralogical composition of the final ceramic product and thus up to a considerable extent the physical properties of the ceramic body. Mineralogical examination of the ceramic body, for example by X-ray Powder Diffraction (XRPD), provides information about the remaining content of phyllosilicate crystals as well as the content of high-temperature crystalline phases formed during firing (Section 2.2; Montana Reference Montana2020). In combination with elemental analysis this allows for determining the type of raw materials and particularly phyllosilicates, such as illites, serpentines, or kaolinites, which had been selected initially for the manufacture of the ceramics under study (Hein & Kilikoglou Reference Hein and Kilikoglou2020a).

3.2.2 Clay Paste Preparation

Depending on the granular composition of the raw materials they have to be processed more or less extensively in order to obtain a workable clay paste (Gosselain & Livingstone Smith Reference Gosselain, Livingstone Smith, Livingstone Smith, Bosquet and Martineau2005). For this, they are commonly first crushed or ground and occasionally sieved in order to remove large non-plastic components. The clay can be furthermore refined by suspending and levigating it in water. Afterwards the refined and homogenized clay paste potentially can be tempered with a controlled amount and size distribution of non-plastic components. Even though in the past the use of diverse temper materials has been occasionally ascribed to cultural factors the temper selection in fact appears to be rather related to the vessel function (Bronitsky & Hamer Reference Bronitsky and Hamer1986). After all, temper materials affect the mechanical performance (Kilikoglou et al. Reference Kilikoglou, Vekinis, Maniatis and Day1998; Müller et al. Reference Müller, Vekinis, Day and Kilikoglou2015; Allegretta et al. Reference Allegretta, Eramo, Pinto and Kilikoglou2015) as well as thermal performance (Hein et al. Reference Hein, Müller, Day and Kilikoglou2008; Allegretta et al. Reference Allegretta, Eramo, Pinto and Hein2014) of the ceramic material. Aside from tempering, another option for controlling the workability of the clay paste and the physical properties of the final product is the mixing of two or more clayey raw materials from different sources, thus combining different types of phyllosilicates (Betina Reference 69Betina2019). Eventually, the clay paste can be mixed also with organic materials, which combust during the firing process. Tempering with organic materials was commonly applied in order to generate additional porosity, which might be advantageous for specific intended functions requiring, for example, thermal shock resistance or the suppression of heat transfer for example in furnace structures (Skibo et al. Reference Skibo, Schiffer and Reid1989; Hein et al. Reference Hein, Karatasios, Müller and Kilikoglou2013). The microstructure of the ceramic body can be investigated with thin-section petrography (see Section 2.1.1). Here, non-plastic inclusions can be identified and assessed in view of natural components or intentional temper, providing information about provenance as well as clay paste processing (Quinn Reference Quinn2022).

3.2.3 Vessel Forming

The next stage in the chaîne opératoire is transforming a lump of the processed clay paste into an object with a specific shape, such as a vessel. A decisive technological development in vessel forming was the introduction of a turning base allowing for controlling the rotational symmetry of the vessel during fashioning. The fast turning potter’s wheel, eventually, allowed for fashioning the clay by applying rotary kinetic energy (RKE) (Roux Reference Roux and Hunt2017; Reference Roux and Courty2019; Thér Reference Thér2020). Nevertheless, there are numerous forming techniques without using RKE, which are still practiced nowadays. A vessel can be built, for example, by assembling coils or slabs or the clay paste can be pressed or casted into a mould. Eventually, the forming technique affects the distribution and orientation of inclusions and voids in the clay body and in the fired ceramic body accordingly as well as its compaction and density. In consequence, it affects the thermomechanical performance of the ceramic body. The initial forming technique can be investigated by surface examination or by optical microscopy (OM) of the microfabric of the ceramic body (Courty & Roux Reference Courty and Roux1995; Choleva Reference Choleva2012; Thér Reference Thér2016). The ceramics can be examined non-invasively with X-ray radiography in order to examine their microstructure in view of forming techniques, which, though, still provides only two-dimensional information (Berg Reference Berg2008; Sanger Reference Sanger2016). Full three-dimensional information of the microstructure can be generated by micro X-ray computer tomography (μCT) (Kozatsas et al. Reference Kozatsas, Kotsakis, Sagris and David2018; Gait et al. Reference Gait, Bajnok and Szilágyi2022).

Apart from the intrinsic properties of the ceramic material, though, the vessel performance is affected by its design as well. In this regard design parameters, for example, height, width, wall thickness, and curvature, affect the extensive properties of the vessel, such as weight, volume, stiffness, or heat capacity. The vessel design is decisive for performance and potential damage risk taking into account surface areas presumably exposed to thermomechanical loads, which depend on intended use and application. It can be conventionally examined by measuring the dimensions and generating scaled drawings or taking scaled photographs. In order to generate a digital 3D model this can be complemented by laser-aided profile measurements (Demján et al. Reference Demján, Pavúk and Roosevelt2023) or 3D scanning through photogrammetry, laser scanning, or structured light scanning (Karasik & Smilansky Reference Karasik and Smilansky2008). In the case of closed vessels, though, it is not possible to retrieve full information about wall thickness by 3D scanning. In this case X-ray radiography or X-ray computer tomography can be applied (Berg Reference Berg2008; Karl et al. Reference Karl, Jungblut, Mara, Wittum, Krömker and Martinón-Torres2014).

4.1.1 Testing of Thermomechanical Properties

The thermal and/or mechanical performance of ceramic materials in view of their intrinsic material properties can be investigated with material testing according to international standards. For these tests commonly small series of standard shaped specimens, typically bars, disks, or cylinders, of each ceramic material under question have to be prepared. The actual testing, though, often implicates the destruction of the ceramic specimen. Thus, instead of specimens cut and prepared from original archaeological pottery alternatively laboratory specimens can be manufactured and tested replicating specific ceramic fabrics taking into account studies of potential raw materials, observed temper materials and estimated firing conditions. Another advantage of replicates concerns potential alteration or pre-existing flaws in archaeological fragments, which might impair their initial thermomechanical performance.

Since material properties depend up to a large extent on the density or porosity of the material, these parameters should be determined as well. The apparent density can be simply determined by dividing the dry mass of a standard shaped specimen by its volume. The pore size distribution and total porosity can be estimated based on micrographs or μCT scans, or it can be measured by mercury porosimetry (Hein et al. Reference Hein, Karatasios, Müller and Kilikoglou2013).

5.1.1 General Considerations

Luminescence dating methods encompass a range of techniques, which are capable of determining the age of crystalline and inorganic, semiconducting, non-metal materials such as pottery and kilns. These techniques are based on the radiation-induced charge population within the crystalline minerals, and record the time since the last event when the charge population was reset (Aitken Reference Aitken1985); this is a physical quantity that increases monotonically with time. Samples for luminescence dating can be divided into two groups. The first group includes geological sediments whose grains have been exposed to sunlight during transport or deposition, along with lithic archaeological artefacts that could be dated using surface exposure dating, such as masonry, megalithic buildings and constructions, monolithic structures and monuments, rock art and mortars (Liritzis Reference Liritzis2011). The second group includes archaeological samples that consist of clayey materials being heated, burnt, fired, or baked in the past. This group includes ceramics, pottery, porcelain, bricks, kilns, and, in general, fired or even burnt materials, along with solidified lava, volcanic, and siliceous rocks (cherts). In that group, which also includes heated lithics, fire-cracked and fire-modified rocks, heating is the zeroing mechanism, and thus, luminescence can date the last heating event; either accidental or intentional.

All crystalline and inorganic, semiconducting, non-metal minerals are exposed to environmental ionizing radiation ever since they were formed. Radiation causes ionization of atoms in these materials, thus creating free electrons and positive ions simultaneously. The same minerals include a number of either impurities or defects, the distribution of which is such that these act as metastable traps for free electrons inside the main matrix of the mineral. Thus, electrons diffuse through the mineral until they finally get trapped in specific defects of the crystal lattice termed traps. The term ‘metastable’ is used in order to assign a trapping lifetime to each one of these traps; unless the mineral is externally stimulated, the electrons could stay trapped as long as tenfold the corresponding trap’s lifetime. In the case of stimulation, the released electrons once again diffuse through the material until they find positive ions (holes) in the lattice and recombine. Each recombination results in the emission of one photon in the optical wavelength band (400–700 nm). This faint light emitted is called stimulated luminescence. Stimulation can take place either inside the lab or during the overall life cycle of the artefact to be dated. The two main stimulation modes include either heating or the action of light (bleaching). Inside the laboratory, in the former case we have thermoluminescence (TL hereafter) while in the latter optically stimulated luminescence (OSL). While OSL is frequently used for dating materials that are involved within the first group (namely geological dating of sediments and rocks), TL is the most appropriate dating technique for the constituents of the second group.

Focusing now on TL, we shall continue by noticing that it stands as the most effective and well-established technique towards both age assessment of heated/burnt materials as well as authenticity testing of pottery, terracotta, and porcelain (Aitken Reference Aitken1985; Wagner Reference Wagner1998). It is important to emphasize that in general, heating to an adequately high temperature is required in order to have an effective zeroing mechanism and thus reliable ages. This temperature should not be lower than 400–450°C. Nowadays, for dating all these heated materials, TL signal from silt particles is highly recommended; alternatively, the OSL signal of sand-sized quartz is also being used. This section will focus on the application of solely TL for dating ceramics, pottery and in general every clayey material being heated, burnt, fired, or baked in the past.

5.1.2 Thermoluminescence Phenomenon and Main Rationale for Dating

Baked artefacts such as pottery, kilns, ceramics, and sherds include a mixture of fine-grained clays and coarser-grained siliciclastic materials, such as quartz and feldspars (Duller Reference Duller2008). Pottery is made by firing clays, which contain 40–60% silica. Most of it is in the crystalline form of quartz; nevertheless, depending on the origin of the clay, feldspar is also an important constituent mineral. These two naturally occurring minerals contain a plethora of traps, among which, and without any external stimulation, some yield lifetimes so short that these are not useful for dating applications. Nevertheless, some could be stable enough to store electrons for extremely long time intervals, depending upon specific physical characteristics of the material. Consequently, thermoluminescence dating is based on the fact that naturally occurring minerals like quartz and feldspars act as natural dosimeters and preserve a record of irradiation energy received through time.

In order to quantify this deposited energy, the concept of accumulated dose (in units of Gy), namely energy per unit mass, is used. Excluding radiation accidents and artificial irradiation because of everyday applications, minerals in general are irradiated by (a) decay of natural radionuclides that exist in their environment of deposition, (b) decay of natural radionuclides within the materials themselves, and finally (c) gamma radiation from cosmic rays. For both decays inside and in the surrounding of the materials, this dose results mainly from the decay of natural radionuclides such as ²³²Th, ⁴⁰K, ⁸⁷Rb, and natural U (both ²³⁵U and ²³⁸U), along with cosmic rays. All these provide a source of low-level ionizing radiation through time that could be considered stable when compared to the overall irradiation time.

The number of trapped electrons is increasing as long as the material is irradiated, without being stimulated. The total amount of trapped charge is proportional to the total radiation energy absorbed by the materials, and therefore to the time subjected to irradiation. The intensity of the luminescence light is also proportional to the total radiation energy absorbed by the materials, and therefore to the time subjected to irradiation (Aitken Reference Aitken1985, Reference Aitken1990). However, for the case of any archaeological artefact, every time that this is subjected to prolonged heating, as in the case of firing, electrons are abruptly evicted and traps are totally emptied; thus, TL signal is totally zeroed. In that case, the material is said to be totally reset. Afterwards, it could start accumulating energy in the form of trapped electrons in order to refill the empty traps once again. The total number of trapped electrons forms a luminescent ‘clock’ which starts measuring time from the beginning (t = 0) every time that these traps are totally emptied. Thus, the last heating signifies the zeroing event of the ‘time-clock’, while the duration of irradiation since the last zeroing effect is associated with the age since then.

Towards the direction of TL age determination, two different physical quantities are required: the total accumulated dose over the past, signifying the total amount of energy deposited inside the matrix, termed paleodose, as well as the rate at which this energy-dose is accumulated, termed dose rate (DR). To be more precise, the concept of equivalent dose is used (D_e, in Gy units), standing as the amount of irradiation dose from either β or γ radiation, which is needed in order to obtain a luminescence signal identical to that resulting from natural radiation exposure. This is a physical quantity that increases monotonically with time and can be measured precisely. The ratio of these two quantities, that is, the equivalent dose over the dose-rate, represents the age of the sample.

$Age\left(kyrs\right)=\frac{EquivalentDose}{DoseRate}=\frac{D_e\left(Gy\right)}{DR \left(\frac{Gy}{kyrs}\right)}$

5.1.3 Sampling Guidelines

As the materials in question are quite compact, it is not mandatory to keep them in the dark before sampling. Nevertheless, samples should be sealed in an airtight plastic bag as soon as they are excavated. Most laboratories require sampling of thick samples, with sizes ≥ 10 mm thick and 30–40 mm across (Nelson et al. Reference Nelson, Gray and Johnson2015). Generally, larger-sized samples result in ages with better precision. When it is possible, multiple sherds/parts of the same object should be collected so that statistics will result in decreased error and uncertainty; this becomes feasible when kilns are to be dated (Aidona et al. Reference Aidona, Spassov and Kondopoulou2021). Finally, in many cases drilling could also be considered as an alternative sampling strategy. In these cases, it is suggested that the drill should be at least 50 mm in diameter while sampling should take place to depths of 100 mm. This will provide an adequate length after discarding the outer 1–2 mm of the core affected by the drilling. This strategy follows on directly from Fleming (Reference Fleming1979), who had suggested drilling as sampling strategy for authenticity testing in pottery; drilling was, and is still supposed to take place from a surface underneath the artefact, at 10 mm in diameter and to depths of 10 mm.

Very fine silt clay yields grains with diameter less than 20 μm while coarse-grained sand includes grains with dimension over 90 μm. The use of small samples (including some hundreds of grains) and single-grain dating of purified coarse grains of either quartz or feldspar have several advantages over poly-mineral fine-grained dating (4–12 μm); the former is routinely used in OSL dating of sediments, while the latter is the standard procedure for TL dating of previously heated materials. After collecting those, the samples are treated inside the laboratory strictly under red dim light conditions. The outer 2- to 4-mm-thick outer layer of the sample should be mechanically removed, in order to eliminate the light-subjected portions (Fleming Reference Fleming1979; Aitken Reference Aitken1985; Reference Aitken1990). The exact thickness to be removed depends strongly on the colour of the sample and thus its opacity; the darker the colour, the less sample should be removed. Application of standard chemical procedures is suggested when silt material is preferred (for an example the readers could refer to Vieillevigne et al. Reference Vieillevigne, Guibert and Bechtel2007). The final step includes gently crushing the sample in an agate pestle and mortar. Finally, both extraction of grains within the grain size range 4–12 μm and final homogeneous precipitation onto 1-cm-diameter aluminium discs is performed by suspension in acetone as the solvent evaporates according to the settling procedure described by Zimmerman (Reference Zimmerman1971). The sample on each disc is termed an aliquot.

The sample’s dose rate could be assessed by calculating the magnitude of the accumulated radiation dose that contributes to the material itself, from factors such as cosmic rays and radioactive elements from the surrounding soil. Thus, sampling should also include the soil closely surrounding the artefact, enabling the measurement of both gamma dose rate and water content. Even though this soil sample also need not be kept in the dark, it should be sealed in an airtight plastic bag as soon as from the first moments of the excavation so that the water content is preserved. Information on the latitude, longitude, and elevation of the site along with the burial depth of the sample is also mandatory so that the cosmic ray contribution to the dose rate could be calculated. In cases where samples must be collected near a context boundary, such as an infill and the natural soil, in situ measurements of the gamma dose rate are strongly suggested (Duller Reference Duller2008). This could be achieved by using either a portable gamma spectrometer or capsule TL dosimeters.

5.1.4 Measuring Thermoluminescence: The Concept of the TL Glow Curve and the Connection to the Traps

While measuring TL, heating is conducted from room temperature up to usually 500°C, with a stable, low heating rate (≤5°C/s). The intensity of the signal is recorded versus temperature and is so weak that it could be measurable only by means of a sufficiently sensitive device called photomultiplier. The temperature dependence of the emission appears as a set of peaks with each one corresponding to an electron-trapping defect; this is the definition of the TL glow curve. TL glow curves in some cases consist of well-defined peaks and in others, of complex sets of overlapping peaks. Each TL peak corresponds to an electron trap of the material under study. TL peaks at low temperatures correspond to shallow electron traps that are unstable with low lifetimes as they can be excited easily. TL peaks at high temperatures correspond to deeper traps which are considered as stable, as they yield lifetimes exceeding some thousands or even hundreds of thousands of years. For a second heating of the same sample without further irradiation, the emission curve consists only of black body radiation and is characterized as a background measurement.

5.1.5 Measurements for TL Dating inside the Laboratory

5.1.6 Data Analysis inside the Laboratory

For each pottery sample, discs are irradiated in groups of four or five at each dose; thus, reproducibility of at least three different TL glow curves for each group is highly recommended. Figure 6 presents typical as well as representative MAAD TL analysis results for a kiln sample reported by Aidona et al. (Reference Aidona, Polymeris and Camps2018). The plot in Figure 6A presents typical additive dose TL curves, each one being the mean value of at least three independently measured glow curves; additive doses are presented in the plot. Error bars indicate excellent reproducibility among the glow curves of the same group-dose. It is highly recommended that the TL signal presents a linear response over increasing doses. Such linearity should be strongly established; this could be achieved by calculating the dose–response slope (S) throughout the entire TL glow curve and subsequently plot it versus temperature for each one of the different additive doses attributed. This slope is calculated according to the following formula (Liritzis et al. Reference Liritzis, Aravantinos and Polymeris2015; Aidona et al. Reference Aidona, Polymeris and Camps2018):

$S=\frac{\left[\left(NTL+{\beta }_i\right)-\left(NTL\right)\right]}{{\beta }_i}\left(\frac{TLCounts}{Gy}\right)$

where (NTL + β_i) denotes the corresponding TL glow curves after each additive dose (β_i) has been administered. The preceding subtraction takes place for each data point. A single individual S curve corresponding to each additive dose is yielded. In the case of linearity, these individual S curves should coincide over an extended temperature region of the glow curve, as the case presented in Figure 6B. This stands among the most important criterion for selecting the appropriate signal integration region of interest for age calculation. The inset of this plot presents a representative example of additive (squares) and second glow (diamonds) TL build up curves for the temperature of 250°C. Equivalent doses are estimated, in units of Gy, according to the following equation:

$D_e= \frac{NTL}{S}\left(\frac{TL Counts}{\frac{TL Counts}{Gy}}\right)$

Once again, the preceding calculation takes place for each data point, over the entire TL glow curve temperature region. For prolonged heating, the plots of equivalent dose versus temperature are expected to yield wide plateaus, with the equivalent dose final values being obtained as the mean value over the plateaus for each sample. A typical plot of D_e against glow curve temperature is presented in Figure 6C. Supra-linearity correction, I, is equal to zero if the second glow TL dose–response passes from the origin of both axes; otherwise, this non-zero value is added to the equivalent dose.

Figure 6 Plot (A) presents examples of TL glow curves of the additive dose procedure, namely natural TL along with three additive glow curves (5, 10, and 15 Gy, respectively). Reheats have been subtracted. Each TL glow curve is expressed in terms of the average over at least three individual TL glow curves. Plot (B) presents the dose-response-slope curve S of TL versus temperature, for the three additive doses applied. Inset shows additive dose growth curves for the temperature of 320°C, for both first (squares) and second glow (points) measurements. The arrow indicates the D_e value. Supra-linearity correction I is estimated to be zero. Plot (C) presents D_e versus temperature, indicating a long plateau

(Reproduced from Aidona et al. Reference Aidona, Polymeris and Camps2018; their figure 7).

5.1.7 Errors, Error Sources, and Limitations

Estimation of accuracy is a particularly tedious aspect in TL dating of pottery. The first important error source is possible poor reproducibility in mass among the aliquots. As a large number of aliquots are being used for getting one single age, reproducibility becomes quite important. Moreover, a dozen or so measured quantities enter into the corresponding calculation, each one among them incorporating the corresponding uncertainty. Thus, the overall uncertainty should reflect the combined effect of uncertainties in these, appropriately weighted (Aitken Reference Aitken1990). Errors are also derived from the uncertainties in curve fitting, especially when lack of linearity is monitored. Whereas the overall random errors on a date can be reduced by averaging the results on a number of synchronous samples, this is quite difficult for the systematic uncertainties. Based on all aforementioned, the latter is the ultimate barrier to better accuracy, presently being of the order ± 7–10% at the 68% confidence level (±1σ).

TL dating can reach back to the earliest pottery and even beyond; how much beyond depends on both properties and origin of the mineral being used. TL ages of baked clay around 30 to 50 kyrs could be easily achieved, while the technique has reported ages as old as 300 kyrs before the current era. At the recent end of the age range limitation, pottery and baked clay objects that are few hundred years old can be safely dated using the previously described procedure, always depending on the levels of both radioactivity and TL sensitivity.

5.1.8 Ages and Interpretation

For the majority of pottery and ceramic items, last heating coincides with the heating while manufacturing the artefact. For ceramics belonging to an archaeological site, dating obtained by scientific techniques is used as a terminus post-quem and combined with the stratigraphic sequence, in order to frame the life of the excavated site into an absolute chronological range. Nevertheless, studies on ceramics are usually wide enough, dealing with the investigation of their overall life cycle, from the acquisition and processing of raw materials to production and distribution, through their use, possible reuse, and discard, along with interpretation in terms of the people producing, distributing, and using them (Galli et al. Reference Galli, Sibilia and Martini2020). This case of discard becomes quite important for kilns, where last heating is closely related to the last time of labour. A very interesting case includes furnaces that had been used for smelting metals (Nerantzis et al. Reference Nerantzis, Kazakis and Sfampa2017); furnace fragments and refractories of severely fired clay, which are usually present in such production sites, represent an ideal material for dating the metallurgical activity.

Nevertheless, TL is the unique dating technique that can identify possible reuse of ceramic items. A very interesting feature is yielded while estimating the equivalent dose values, indicating two individual plateaus in the corresponding plots versus temperature. The secondary plateau is observed at higher temperatures than that of the primary and main plateau, corresponding at the same time to a larger equivalent dose and thus to an older age. The presence of two plateaus resulting for the same sample identifies two independent zeroing events in the past, in conjunction to insufficient heating of the sample recently, namely heating at firing temperatures around or lower than 500°C. The second, milder zeroing event due to insufficient heating is usually interpreted as either a fire event in recent history or a reuse period of the artefact (Liritzis et al. Reference Liritzis, Aravantinos and Polymeris2015).

When possible, the application of more dating techniques, together with the available archaeological evidence, offers the best approach for obtaining a more precise chronological framework for an archaeological site. Archaeomagnetic and TL dating techniques share the same rationale, dating exactly the same event that is the last heating of the artefacts baked at high temperature. During the last years, the archaeomagnetic method is considered as complementary dating tool for baked clay to TL, not only in cases where other more traditional dating methods failed to estimate the age of the structures but also due to its capacity to date directly the structures themselves and not findings they contain (Kontopoulou et al. Reference Kontopoulou, Aidona, Ioannidis, Polymeris and Tsolakis2015; Aidona et al. Reference Aidona, Polymeris and Camps2018, Reference Aidona, Spassov and Kondopoulou2021).

5.2.1 Mineralogy of Quartz

As pottery contains 40–60% silica, most of it being in the crystalline form, the presence of quartz is ubiquitous, in all ceramic samples. Given that it is one of the most abundant minerals occurring at the Earth’s crust, it is not surprising that it generally represents the prevalent phase in ceramics. At ambient temperatures, α-quartz crystallizes to a trigonal polymorph and is stable up to 573°C; this is the temperature threshold for the transition of α-quartz to β-quartz. This latter phase crystallizes in a hexagonal form and is also stable. In fact, this transition is reversible involving a volume decrease and, during cooling, β-quartz transforms back into α-quartz (Preusser et al. Reference Preusser, Chithambo and Götte2010). If the temperature reaches 872°C, β-quartz transforms into HP-tridymite which later on transforms into β-cristobalite at 1470°C (Heaney Reference Heaney, Heaney, Prewitt and Gibbs1994). Transition from β-quartz to tridymite is irreversible and thus tridymite cannot transform back to β-quartz. However, the tridymite transformation is bypassed when firing pure quartz; in this case, in fact, β-cristobalite is directly formed at ~ 1050°C, and this transition is reversible (Heaney Reference Heaney, Heaney, Prewitt and Gibbs1994; Gliozzo Reference Gliozzo2020). Quartz appears to be stable up to high temperatures of about 850–900°C, and as it proceeds towards higher temperatures, its amounts can gradually or abruptly decrease. The mineralogy of quartz can provide useful firing temperature thresholds, based on the detected quartz polymorphs, the corresponding transition temperatures, and the formation of new phases depending on the surrounding established minerals. Nevertheless, the following sections will describe an alternative and innovative experimental approach for the assessment of firing temperature of pottery, presenting a methodology that takes full advantage of TL properties of quartz (Sunta & David Reference Sunta and David1982). For other techniques towards assessment of firing temperature in the past, either mainstream or alternative, the readers could refer to Gliozzo (Reference Gliozzo2020).

5.2.2 Rationale

Quartz mineral, besides being a good dosimeter-chronometer, ubiquitous, and resistant to weathering, is the most widely used mineral for TL that exhibits some quite interesting properties: (a) under the combined action of irradiation and heating, the sensitivity of the TL signal (namely the quantity of TL per unit mass and dose) does not remain constant but increases instead; this is the so-called pre-dose sensitization (Zimmerman Reference Zimmerman1971; Bailliff Reference Bailiff1994). This effect involves simultaneous sensitization due to both thermal and dose contribution to the 110°C TL peak pre-dose method (Bailiff Reference Bailiff1994); (b) it can retain information regarding its thermal history in terms of the maximum temperature attained (Sanjurjo-Sanchez et al. Reference Sanjurjo-Sanchez, Montero Fenollos and Polymeris2018). These unique as well as prevalent properties of all quartz samples all around Earth have been exploited in the past to assess the firing temperature of pottery. While measuring TL in quartz, the corresponding glow curve shows mainly four main peaks with maximum intensities at 110°C, 200°C, 325°C, and 375°C (at moderate heating rates; Franklin et al. Reference Franklin, Prescott and Scholefield1995). The 110°C TL peak, being correlated to a shallow trap, is useless for dating. Nevertheless, the luminescence characteristics of the 110°C peak can be directly correlated to the thermal history of quartz and specifically to the firing temperature (Watson & Aitken Reference Aitken1985).

The sensitization of the 110°C TL peak of quartz occurs within the temperature range over 200°C, provided that this temperature does not exceed 880°C where the phase transition of β-quartz (to tridymite) is irreversible. Higher temperatures can cause vitrification and high-temperature minerals that easily allow identifying the firing temperature, mostly using SEM. Once a pottery sample is re-annealed (or re-fired) in the laboratory at a temperature that is higher than the firing temperature of the original pottery sample (where from quartz was extracted) sensitization of the 110°C peak’s TL intensity is observed. The underlying idea of the 110°C TL peak sensitization method is that thermally induced hole transfer of charge associated with geological irradiation takes place during firing, and re-firing experiments can reconstruct the point at which additional sensitization takes place. Even though the signal of the high-temperature TL peaks could also be effectively used for thermometry (Spencer & Sanderson Reference Spencer and Sanderson2012; Oniya et al. Reference Oniya, Polymeris, Tsirliganis and Kitis2012; Polymeris et al. Reference Polymeris, Kiyak, Koul and Kitis2014), especially for temperatures over 600°C, the 110°C TL peak is suggested as the best probe. From a practical point of view, the use of this specific signal is quite useful, as this is a single, ubiquitous, and prominent peak that does not require deconvolution analysis.

However, there are a number of requirements for such behaviour to work – including the need for thermal stable hole reservoirs (from which the initial firing can transfer charge carriers) in sufficient supply that they do not saturate under geological conditions, and that the luminescence centres should be thermally stable over archaeological timescales so that such information is preserved. These requirements are not always fulfilled while the conditions required for effective application cannot be assured in advance.

5.2.3 Sampling, Handling, and TL Measurements

For palaeothermometry applications, the most external layer of the pottery samples should be removed in order to remove possible corrosion and oxidation products. About 5 gr of each sample should be scraped off and gently crushed. For TL analyses, use of 90–250 μm grain-size is suggested to get enough quartz amounts, whereas the 90–180 μm grain-size fraction is recommended (Polymeris et al. Reference Polymeris, Kiyak, Koul and Kitis2014).

Towards obtaining pure coarse quartz extracts, the samples are sieved to get the 90–250 μm grain-size fraction; grains are water-washed and treated with HCl and H₂O₂ to remove carbonates and any remaining organic matter, respectively. Heavy liquids of densities 2.62 and 2.70 g/cm³ should be used to separate feldspars, quartz, and heavy minerals. HF application to the quartz extract removes any feldspar grains, and HCl is applied again after water-washing to remove any remaining soluble fluorides.

Initially, each pottery sample should be divided into a number of segments. Each segment is to be annealed (or re-fired) at a different temperature, in steps of 50°C. The selection of the appropriate temperature region depends strongly on the samples to be studied, with the prerequisite that should bracket the original firing temperature. Preliminary mineralogical analysis could constrain this aforementioned temperature range. Annealing can be performed in external automatic furnaces. The annealing duration in all cases should be at least 1 hour in an oxygen atmosphere; this is the typical firing duration for pottery. Independent of the temperature, in all cycles, the heating that is applied inside the laboratory is termed re-firing.

Historically, the first protocol that was experimentally applied was the protocol of Thermal Activation Curves (TAC) by Göksu et al. (Reference Göksu, Wieser and Regulla1989) in TL. Subsequent work has demonstrated that the full pre-dose protocol suggested by Polymeris et al. (Reference Polymeris, Kiyak, Koul and Kitis2014 and references therein) is much more robust and was successfully used to assess the firing temperature of pottery samples. This protocol has also been proven reliable for fired stones (Sanjurjo-Sánchez et al. Reference Sanjurjo-Sánchez, Gómez-Heras and Polymeris2013). This protocol includes three sequential cycles of dose and TL measurements, designed to monitor the dependence of both sensitivity and sensitization of the 110°C TL peak due to (a) the thermal contribution of the pre-dose effect (Koul et al. Reference Koul, Chougaonkar and Polymeris2010) and (b) the full pre-dose effect to gradually increasing heating. Each TL measurement should be performed at low heating rate, up to a maximum temperature of 500°C. Figure 7 presents an example of TL glow curves measured in the framework of this optimized protocol up to 500°C, following a selection of re-firing temperatures. This figure presents fully pre-dosed TL glow curves that were measured during the third cycle of dose and measurement. For more details, the readers could refer to protocol (A) from Polymeris et al. (Reference Polymeris, Kiyak, Koul and Kitis2014).

Figure 7 Example of pre-dosed TL glow curves for quartz extracted from Mesopotamian pottery samples; each curve is the average of three independent measurements. Arithmetic indicates the re-firing temperature

(Reproduced from Polymeris et al. Reference Polymeris, Kiyak, Koul and Kitis2014; their figure 1).

The working principle includes studying the temperature-dependent changes of both thermal and pre-dose sensitization of the 110°C TL peak in quartz due to re-firing. The plot of the “110 °C TL peak” sensitization versus re-firing temperature, normalized over the respective sensitivity at the lowest re-firing temperature is a very useful tool. No sensitization takes place at low re-firing temperatures, as the normalized sensitivity of the probe peak gets values around 1. As long as this latter temperature is lower than (or even equal to) the firing temperature in the past, the intensity of the 110°C TL peak doesn’t change. By the time that re-firing temperature in the lab becomes higher than the firing temperature, re-firing temperature dominates due to being higher than the temperature of the maximum heating, inducing sensitization; thus, a prominent and steep increase in both sensitivity and sensitization is monitored. At these temperatures, a prominent spike is observed when this ratio is plotted versus re-firing temperature. However, for this latter plot, typical experimental features, such as the monotonic increase of the sensitization ratio above the annealing temperature along with a decrease at re-firing temperature around 900°C, are expected (Koul Reference Koul2006; Polymeris et al. Reference Polymeris, Sakalis and Papadopoulou2007, Reference Polymeris, Kiyak, Koul and Kitis2014).

All these features are obvious in Figure 8 that presents the results of the TL analysis in terms of the normalized 110°C TL versus the initial re-firing temperature obtained for one ancient ceramic coffin of a Japanese Emperor (Tema et al. Reference Tema, Hatakeyama and Ferrara2024). The TL sensitivity of the 110°C TL peak does not depend on the laboratory re-firing temperature up to 525°C, representing the last re-firing temperature that indicates the last stable (normalized) sensitivity. As re-firing temperature in the lab increases further, all three sensitivities show a steep increase, with the one due to pre-dose sensitization being the most prominent. This increase of sensitivity is not linear versus the re-firing temperature, which is a typical feature of firing temperature below 550°C. In such cases, the maximum heating temperature is considered as the temperature just before this prominent and steep sensitivity change, estimating in this case firing temperature of around 525±25°C; an error of ±25°C is considered as the smallest re-firing temperature step used in the luminescence protocol.

Figure 8 Normalized TL sensitivity versus laboratory re-firing temperatures for the 110°C TL peak measured from one sample of an ancient ceramic coffin of a Japanese emperor. Normalization has been carried out over the corresponding first data point. The plots represent re-firing sensitization (squares), thermal sensitization (dots), and pre-dose sensitization (diamonds). Re-firing, thermal, and pre-dose sensitizations are recorded at various re-firing temperatures in the range between 300°C and 750°C in steps of 50°C

(Reproduced from E. Tema et al. Reference Tema, Hatakeyama and Ferrara2024; their figure 4).

6.3.1 Chemical Composition of the Glaze

Glazes are composed of three main components: silica, alumina, and a flux. Silica is the main glass former, alumina provides stability and prevents excessive flow, and a flux is an element that lowers the melting point of the glaze and helps it adhere to the ceramic body. Glazes may also contain various additives, such as colourants, opacifiers, and modifiers, to achieve different aesthetic and technical results.Footnote ⁴ Glazes can be classified according to their firing temperature, which affects their composition and properties. Low-temperature glazes are fired at temperatures between 900°C and 1050°C, while high-temperature glazes are fired at temperatures above 1100°C. The fluxes of low temperature glazes are lead, sodium, and potassium oxides, while the fluxes of high-temperature glazes are calcium, sodium, and potassium oxides. Frits are pre-melted glaze ingredients that are ground into a powder and mixed with water before they are applied to the ceramic.

6.3.2 Opacifiers

Transparency in ceramic glazes is not always desirable, as the colour of the underlying ceramic body (from buff colour to red or black) can interfere with the colour of the glaze and decorative motifs. Applying a white slip or the bleaching of the ceramic surface were common early methods of avoiding this, but the production of opaque glazes was soon adopted. Opacity in glazes can be achieved by the generation of a large number of bubbles, by the presence of undissolved quartz or feldspar particles, or by the presence of small precipitates (about 1 µm) of cassiterite, SnO₂, wollastonite, pyroxene or feldspar, dispersed in the glaze, or, as mentioned earlier, precipitate at the glaze-ceramic interface, with net crystalline surface which scatter light more efficiently increasing the opacity (Figure 11). Large bubbles and undissolved particles produce a partially opaque glaze, while the small crystallites formed during the glaze firing with neat crystalline surfaces and a high refractive index further increase the opacity. In particular, the addition of an excess of quartz particles to the glaze mixture results in a glaze full of undissolved quartz particles which act as a white opacifier (Figure 12). This method has been used in particular in those periods and regions (9th- and 10th-century Sicily and North Africa) (Salinas et al. Reference Salinas and Pradell2020; Salinas et al. Reference Salinas, Reynolds and Pradell2022) that did not have other opacifiers, in particular, as we will see next, tin. Tin is not as common as lead, and it is known to have been traded from remote regions (Uzbekistan, England) to the Mediterranean production centres.

Figure 12 Fatimid polychrome ceramic from Tunisia, 10th century (Bir Ftouha, BFA47) (Salinas et al. Reference Salinas and Pradell2020), (A) OM image of a cross section of an overglaze brown decoration. (B) Image of the fragment. (C) BSD image of the same cross section. (D) Dark-field and (E) bright-field OM images of an enlarged area. (F) Enlarged area of E. (G) Enlarged area of F.

When a mixture of PbO, SnO₂ and SiO₂ is fired together, it undergoes several transformations resulting in the formation of yellow lead-tin oxide (Pb₂SnO₄) particles, which then transform to (Pb,Si)₂Sn₂O₆ particles at higher temperatures, and finally dissolve in the melt to recrystallize as well shaped SnO₂ particles, usually less than one micron in size. The temperatures at which these transformations occur depend on the composition of the mixture, with higher PbO/SiO₂ ratios resulting in more stable (Pb,Si)₂Sn₂O₆ particles. Therefore, yellow glazes typically have a high PbO/SiO₂ ratio (typically 62–68% PbO and 24–30% SiO₂) (Figure 11 G-J) (Molera et al. Reference Molera, Pradell, Salvado and Vendrell-Saz1999; Tite, et al. Reference Tite2008; Tite et al. Reference Tite, Watson and Pradell2015). In order to decrease the lead content, plant ashes were added in the white glazes to reduce the PbO content and increase other fluxes as Na, K and Ca (Figure 11 D-F; Matin et al. Reference Matin, Tite and Watson2018).

Tin dissolves in alkaline glazes, but can appear precipitate at the glaze-ceramic interface as a calcium tin silicate, malayaite, CaSn(SiO4)O (Tite et al. Reference Tite, Shortland and Paynter2002). However, as a reaction compound, it is not uncommon in copper glasses and glazes, as tin is a common impurity associated with copper. Similarly, brizziite, NaSbO3, has also been found in some alkaline glasses and glazes when antimony is present (Amadori et al. Reference Amadori, Matin and Poldi2023).

Opacity can also be achieved in lime glazes (high temperature) when the SiO₂/Al₂O₃ ratio is high (above 6–7). In this case, the high temperature immiscibility between a lime-rich glass and a silica-rich glass results in a glaze nanostructure. Typical structures are lime-rich droplets in a silica-rich matrix, or silica-rich droplets in a lime-rich matrix, or mixed nano-structures (Figure 11A-C). Light scattering of these structures is responsible for the opacity of the glazes. In modern times (19th century) titanium oxide has replaced tin oxide as a white opacifier.

6.3.3 Colourants and Decorations

Colour is usually added to glazes by the addition of metal cations, among which the most common are Mn²⁺, Mn³⁺, Co²⁺, Cu⁺, Cu²⁺, Fe³⁺, and Fe²⁺. Often, there is more than one present, and a balance is achieved between the different species present, which gives the glaze its characteristic colour. They can also appear in different site coordinations, four-, five-, or six-fold, and this also determines the colour of the glaze (Weyl Reference Weyl2016).

Metal cations can also react with the glaze components during the firing to form crystalline compounds. The reaction compounds formed depend on the composition of the glaze, the firing temperature, the atmosphere, and the concentration of the metal itself. Consequently, dissolved transition metal cations and crystalline precipitates are characteristic of the materials, method of application and firing process.

Colour can be applied to the entire glaze by mixing the colourants with the raw glaze, or decorative motifs can be painted. Decorations are particularly interesting because they can be made in a wide variety of ways: overglaze, underglaze, enamel, side-by-side, cuerda seca, with one or more firings. Depending on the process used different crystalline compounds are obtained, which appear heterogeneously distributed in the glaze. Underglaze painting is applied to a raw dry ceramic surface, followed by a first firing, and then the ceramic is glazed and subjected to a second firing (Figure 13). This allows finer designs to be drawn while the first firing fixes them and reduces blurring. Although it is also possible to use a single firing, the designs will be less defined. Overglaze decorations are painted directly over the raw glaze applied either over a biscuit or raw ceramic which is then fired (Figure 13). Painting over a raw glaze is more difficult which, together with the diffusion of the metal cations, makes the designs less precise. The pigments are often mixed with some clay or raw glaze mixture to facilitate the reaction with the ceramic/glaze and to fix them. Side-by-side painting refers to the application of different coloured glazes directly over the ceramic surface. Enamels are a prepared mixture of glass with a pigment or colourant that is fired at a lower temperature onto the glazed object. This is very commonly used in porcelain and stoneware, less commonly, although not unknown, in low-temperature ceramics. Finally, the cuerda seca, a brown or black paint made from manganese and/or iron oxides mixed with a greasy compound, was used to delineate and separate the areas to be given different colours. The greasy compound spits out the colours, preventing them from mixing.

Figure 13 (A) Image of an Aglabid Raqqada type, polychrome ceramic from Tunisia, 9th century (CS66) (Salinas et al. Reference Salinas and Pradell2020). (B) SEM BSD image of a cross section of the brown area. (C) Bright-field and (D) dark-field OM images of a detail of the brown decoration.

A detailed study of the reaction compounds and their distribution in the glaze gives direct information about the firing conditions (temperature and atmosphere) and the decoration method used.