Navigation – Plan du site

AccueilNuméros41-1Determination of the chemical com...

Determination of the chemical composition of medieval glazed pottery from Drastar (Bulgaria) using PIXE/PIGE and LA-ICP-MS

Composition chimique de poterie glaçurée medievale de Drastar (Bulgarie) déterminée par PIXE-PIGE et LA-ICP-MS
Valentina Lyubomirova, Žiga Šmit, Helena Fajfar et Ivelin Kuleff
p. 69-82

Résumés

Quinze échantillons de céramiques glaçurées médiévales de Drastar (aujourd'hui Silistra), en Bulgarie, datés entre les 10ème et 18ème siècles après J.-C. ont été étudiés. Les compositions élémentaires des glaçures ont été déterminées par les méthodes analytiques : émission de rayons X induite par protons (PIXE) et émission de rayons gamma induite par protons (PIGE). La composition de la matrice argileuse a été étudiée par spectrométrie de masse á plasma inductif avec ablation laser (LA-ICP-MS) sur pastilles des poudres d'argiles. En complément, la microscopie électronique à balayage couplée á la spectroscopie X en dispersion d’énergie (SEM-EDX) a été aussi utilisée pour la caractérisation des matrices argileuses.
Après soustraction des teneurs en PbO et CuO, la comparaison entre les compositions des matrices et celles des glaçures a révélé que tous les échantillons (pâtes calcaires et non calcaires) ont été glaçurés en utilisant de l'oxyde de plomb suivi d'une cuisson en atmosphère oxydante. De plus, il a été montré que le fer, le cuivre et les oxydes de manganèse ont été utilisés pour obtenir les couleurs souhaitées.

Haut de page

Texte intégral

The authors would like to express gratitude to Assist. Prof. Dr. Rumyana Koleva for the valuable comments and discussion and also to Assist. Prof. Dr. Chavdar Kirilov from the Faculty of History at the University of Sofia for providing the glazed pottery samples. One of the authors (Valentina Lyubomirova) is thankful to the IAEA-Vienna for funding the fellowship at the Jožef Stefan Institute, University of Ljubljana, Slovenia.

1. Introduction

1From the 4th millennium BC onwards in the Near East region and Egypt alkali glazes were used first for covering steatite and ground quartz vessels (i.e. Egyptian faience) and then clay vessels from the 15th century BC (Tite et al., 1998).

2The earliest known lead glazed ceramics in the West were produced during the Roman era (about 1st century BC) (Tite et al., 1998; Wolf et al., 2003; Walton and Tite, 2010) after their appearance in China perhaps a few centuries earlier (Tite et al., 1998).

3Evidence of the production of these wares comes from several sites in Anatolia (present-day Turkey) (Walton and Tite, 2010). Subsequently, glazes of this type were extensively used for pottery and tiles throughout the Islamic and Byzantine world and medieval Europe (Wolf et al., 2003). At present their use is continuing in the Near East and Europe (Tite et al., 1998; Wolf et al., 2003). From the late 12th century AD, the high lead glaze type was also used in China as the basis for colored overglaze enamels (Tite et al., 1998).

4An important category of glazed pottery, first produced in Iraq in the 8th century AD is the tin-opacified glazes, initially containing up to 1-2% PbO (Charalambous et al., 2010). Gradually the percentage of PbO increased and by the 10th-11th century AD the glazes typically contained 20-40% PbO and 5-12% alkali (Wolf et al., 2003). According to Davis and Stocker (2013) typical concentration of SnO2 in tin-opacified glazes amounts to 4-9 wt.% Thereafter, the produced tin-opacified glazes throughout the Near East and Europe are characterized by this composition.

5The reasons for replacement of the alkali glazes with lead glazes are lower melting temperature, higher fluidity and thus better coating the glazed surfaces. In addition lower thermal expansion which predetermines lower risk of glaze crazing and higher refractive index leading to greater brilliance (Wolf et al., 2003) are factors that led to the widespread use of high-lead and lead-alkali glazes.

6There are mainly two ways to produce high lead glazes. The first one is to use a lead compound, such as litharge (PbO), red lead (Pb3O4), white lead (2PbCO3.Pb(OH)2) or galena (PbS). In the second way, a suspension of a lead compound with silica in the form of quartz sand, ground quartz or chert pebbles is used. In both methods, a small amount of clay, gum or starch can be added to the glaze slurry to maintain the lead and silica particles in suspension. The application can be to either unfired or biscuit fired bodies (Tite et al., 1998). The lead oxide-plus-quartz suspension can be applied either in a raw state or after pre-fritting (Walton and Tite, 2010).

7The lead glazes are easily formed, obtain lustrousness and opacity at low temperature and are also easily colored with oxides of other metals (Charalambous et al., 2010). Apart from the concentration, the oxidation states of copper and iron are also important for obtaining a particular color. The green color is traditionally related to the presence of Cu2+(Davis and Stocker, 2013) or Fe2+ (De Benedeto et al., 2004; Molera et al., 1997; Wakamatsu et al., 1987). If no copper has been detected in the glazes, the change of the color must be related to the degree of oxidation of the iron (Molera et al., 1997; Wakamatsu et al., 1987). Fe3+ is responsible for the brown, yellow, black and red colors (Molera et al., 1997), however, black glazes also have Mn and yellow glazes also contain Cr (Charalambous et al., 2010). Co-based pigments were well known for the blue color (Wolf et al., 2003) due to CoO4 complex (Charalambous et al., 2010) or probably Co-Ni-minerals (Kleinmann, 1991).

8The production of lead glazes expanded rapidly, attributed to the growing Roman markets and since the 7th century AD lead glazes were produced also in the Balkans (Walton and Tite, 2010).

9The production of glazed pottery in Bulgaria is about 2% of the pottery in general. Glazed pottery has been found during archaeological excavations in Drastar and the nearness (Koleva, 2008; Mancheva, 1986), Karaachteke near Varna (Yoleva et al., 2015), Zlatna Livada, near Chirpan (Koleva, 2015), Pliska (Djingova and Kuleff, 1992), etc. Fragments of glazed ceramic vessels although not many, are always part of the ceramic finds from the Early Medieval pits and houses, which lay above or are dug into the Late Antiquity stratigraphic layer. Its fragmentation is a result of the representative character of this valuable ceramics and the fact that its quantity was considerable less in comparison to the other vessels. The fragments allow identifying the major shapes of vessels, which are very diverse – pitchers, handles, dishes, vessels, etc [Koleva, 2009].

10Glazed pottery from Bulgaria has not attracted much interest in terms of analytical studies. The subjects of the present investigation are fifteen medieval glazed pottery vessels. The samples are taken from town Drastar (present day Silistra) (see Fig. 1) where in the recent years medieval kilns for pottery production were discovered during archaeological excavations (Angelova and Koleva, 2005; Koleva, 2008). The excavated kilns are evidence for the production of sgraffito pottery in Drastar in the period from 13th to 14th century AD [Koleva, 2008]. A detailed description of the town Drastar can be found in Angelova (1988).

11The aims of the present paper are to determine: (i) the chemical composition of the pottery; (ii) the type of colorants in the glazes after chemical characterization of the glaze; (iii) the technology of production of high lead glazes from Drastar.

Figure 1: Map of Bulgaria with the location of Silistra / Carte de la Bulgarie avec l'emplacement géographique de Silistra

Figure 1: Map of Bulgaria with the location of         Silistra / Carte de la Bulgarie avec l'emplacement         géographique de Silistra

2. Materials ands methods

Materials

12In the the present investigation fifteen glazed pottery fragments from Drastar were analyzed. Four of the samples (Nrs. from 1 to 4) were found in the room leading to the uncovered kiln. This fact is a further evidence for the existence of medieval kilns for pottery production in the city, as already established by Koleva, 2008.

13The rest of the samples (Nrs. from 5 to 15) were found during excavations of part of the medieval site of Drastar. Two of the samples (Gl-P-10 and Gl-P-11) are probably imported from Byzantium.

14The investigated samples are only fragments (not entire objects) of sgraffito and glazed ceramics which represent parts mainly from vessels and dishes – bottom, edge and side part and one handle. Most of the samples are glazed from one side and two of them are glazed from both sides. The glazes have different color, dominated by green, brown and yellow in different shades laid on a red, grey or white clay.

15A short description of the pottery samples is given in Table 1. Pictures of the samples with numbers from Nr. 5 to Nr. 15 are presented in Figure 2. The samples are dated archaeologically to 10th century (Nr. 10), 11th century (Nr. 8), 10th to 12th century (Nr. 11), 13th century (Nrs. from 1 to 6, and 14), 13th-14th century (Nr. 9), 14th century (Nr. 15), 16th century (Nrs. 7 and 13), 18th century (Nr. 12).

Table 1: Description of the investigated glazed pottery from Drastar (present day Silistra) / Tableau 1 : Description de la poterie glaçurée de Drastar (aujourd'hui Silistra) qui a fait l’objet d’analyses

Lab. code

Description

Gl-P-1

sgraffito pottery, 13th cent. AD, lead glaze before firing with white-yellowish color, the find was discovered in the room which leaded to the kiln

Gl-P-2

sgraffito pottery, 13th cent. AD, lead glaze before firing with white-yellowish color, the find was discovered in the room which leaded to the kiln

Gl-P-3

sgraffito pottery, 13th cent. AD, lead glaze before firing with white-yellowish color, the find was discovered in the room which leaded to the kiln

Gl-P-4

sgraffito pottery, 13th cent. AD, lead glaze before firing with white-yellowish color, the find was discovered in the room which leaded to the kiln

Gl-P-5

sgraffito pottery; 13th cent. AD, bottom of vessel, red clay

Gl-P-6

sgraffito pottery, 13th cent. AD, bottom of vessel, red clay

Gl-P-7

glazed pottery, 16th cent. AD, edge of vessel, red clay

Gl-P-8

glazed pottery, 11th cent. AD, handle, grey clay

Gl-P-9

glazed pottery, 13th to 14th cent. AD, bottom of vessel, glazed from both sides, white clay

Gl-P-10

glazed pottery, 10th cent. AD, bottom of vessel, glazed from both sides, white clay, probably imported from Byzantium

Gl-P-11

sgraffito pottery, 10th to 12th cent. AD, bottom of vessel, white clay, probably imported from Byzantium

Gl-P-12

glazed pottery, 18th cent. AD, side part of vessel, red clay

Gl-P-13

sgraffito potery, 16th cent. AD, side part of dish; red clay

Gl-P-14

sgraffito potery, 13th cent. AD, bottom of vessel, red clay

Gl-P-15

sgraffito potery, 14th cent. AD; bottom of vessel; red clay

16The surface of some glazes suffered severe deterioration due to environmental erosive factors as humidity and soluble salts. This effect is clearly visible on samples Gl-P-10.1, Gl-P-10.2, Gl-P-11 (see Fig. 2). More information about the analyzed samples is given in Koleva, 2008 and Mancheva, 1986.

Figure 2: Pictures of some of the glazed pottery samples / Photos de certains des échantillons de poterie glaçurée

Figure 2: Pictures of some of the glazed           pottery samples / Photos de certains des échantillons de poterie           glaçurée

17The analyzed samples are very common and typical and are chosen as representative samples for the medieval glazed pottery, excavated in the Drastar region.

Analytical methods

Proton induced X-ray and proton induced gamma-ray emissions

18The glaze covered pottery fragments were analyzed using Proton Induced X-ray Emission (PIXE) and Proton Induced Gamma Emission (PIGE). The measurements were performed at the Tandem accelerator of the Jožef Stefan Institute in Ljubljana, Slovenia using a proton beam of 3MeV nominal energy in air. A combined PIXE-PIGE method was applied. The concentration of Al2O3, CaO, CuO, Fe2O3, K2O, MgO, MnO, Na2O, PbO, SiO2, Cr2O3 and TiO2, was determined. Elements heavier than silicon were analysed according to their characteristic X-rays, detected by a Si(Li) detector of 160 eV resolution at 5.89 keV. The proton energy at the target, after passing an 8 μm aluminium window and a 1.1 cm air-gap, was 2.70 MeV. The air gap between the target and X-ray detector was 5.7 cm, which acted as an efficient absorber of intense silicon X-rays. The beam size at the target was a few mm2. Typical measurements times were 300-500 seconds at a proton current of <1 nA. Sensitivity for mid-Z element was improved to about 5 mg kg-1 by an aluminium absorber of 0.1 mm thickness, and increasing the proton current to a few nA. Spectral deconvolution was performed by the AXIL code.

19The contents of Na, Mg and Al were determined from the intensities of gamma rays excited by inelastic proton scattering. As the proton exit window, a 2 μm thick tantalum foil on a brass nozzle was used. Gamma rays were detected by a 40 % intrinsic germanium detector. The gamma lines exploited for analysis were 440 keV for Na, 585 keV for Mg, and 844 and 1014 keV for Al. Line intensities were determined by the GRILS program of the GANAAS software package. For gamma measurement, the proton current was 2-3 nA and the collected dose was about 5 μC for the sample, and 15 μC for the standard.

20The elemental concentrations were calculated by an iterative procedure, using simultaneously both X-ray and gamma-ray intensities. The contents of Na, Mg and Al were determined within the surface approximation, using NIST SRM 610 glass standard for calibration. Periodically the NIST SRM 610 and NIST SRM 612 glass standards were measured as unknown samples. The accuracy of the method was within 5 % for major elements, but decreased to 10-20 % for the elements at the sub 0.1 % level. Uncertainties are also large for Mg approaching the detection limit; at about 0.4 % they may be as high as 30 %.

Laser ablation mass spectrometry with inductively plasma

21The concentration of the elements in pottery was determined by Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICP-MS). Due to the different porosity of the samples which lead to irreproducible results by direct ablation of the pottery, the samples were prepared in the form of pellets. A small part of the surface ceramic body layer was removed by a drill and a tungsten carbide cutter to eliminate possible surface contamination effects. The samples (0.1g) were ground in agate mortar for 20 min, then mixed with H3BO3 in ratio 1:5, well homogenised and ground again. Afterwards the mixture was pressed into a pellet of 1.2 cm diameter and 2 mm thickness under pressure of 2 bars for 1 min. This approach has been developed for ED-XRF analysis of soils (Ivanova et al., 1998). The calibration of the LA-ICP-MS system was performed by the Certified Reference Material Ohio Red Clay, prepared by Harbottle (1976). Pellets of the Certified Reference Material Ohio Red Clay was prepared the same way and used for calibration of the LA-ICP-MS system. A summary of the analytical data for 42 elements obtained by analysing Ohio Red Clay from different laboratories are published in Kuleff and Djingova (1998). The published mean values of the elements of interest were used as certified values for the calibration of the LA-ICP-MS system.

22The LA-ICP-MS conditions for measurement of the elements of interest are given in Table 2A (ICP-MS, Perkin Elmer SCIEX DRC-e) and 2B (LA, New Wave Research). For the simultaneous determination of total element composition in the samples, the signal of the matrix elements was reduced by the introduction of the rejection parameter (RPa) to enable the high-mass cut off. The optimized RPa values for the measured isotopes of the matrix elements are listed in Table 2.

Table 2: LA-ICP-MS operation conditions for the quantitative analysis of the pottery / Tableau 2 : Les conditions de fonctionnement de LA-ICP-MS utilisées pour l'analyse quantitative de la poterie

A. ICP-MS

B. LA

Argon plasma gas flow

15 L min-1

Wavelength

213 nm

Auxiliary gas flow

1.20 L min-1

Laser ablation chamber

Standard


(New Wave Research)

Nebulizer gas flow

0.90 L min-1

Ablation mode

Continuous

Lens voltage

6.00 V

Line length

4000 µm

ICP-RF power

1100 W

Pulse duration

5 ns

Pulse stage voltage

950 V

Beam diameter

100 µm

Dwell time

50 ms

Fluence

7.5 J cm-2

Acquisition mode

Peak hop

Repetition rate

10 Hz

Peak pattern

One point per mass at maximum peak

Scanning speed

10 µm s-1

Number of runs

4

Carrier gas (Ar) flow rate

1.0 L min-1

Determined isotopes (RPa)

23Na (0.013), 27Al (0.015), 28Si (0.016), 42Ca (0.013), 39K (0.016), 54Fe (0.015), 56Fe (0.015), 57Fe (0.013), 24Mg (0.014), 47, 48, 49Ti, 50, 52Cr, 55Mn, 63,65Cu, 207, 208Pb

SEM–EDX microanalyses

23Scanning Electron Microscopy (SEM), (TESCAN, SEM/FIB LYRA I XMU) with Energy Dispersive X-ray Spectroscopy (EDX) (Quantax 200, Bruker) was applied used to obtain surface images of some of the glazes and potteries. Before analysis the samples were coated with graphite.

3. Results

24The analytical data for the composition of the clay bodies of the fragments, determined by LA-ICP-MS are reported in Table 3.

Table 3: Chemical compositions of the pottery (normalized to 100%) from medieval Bulgaria using LA-ICP-MS; the number of the decimal places is related to the precision of the measurements / Tableau 3 : Compositions chimiques de la poterie (normalisées à 100%) de la Bulgarie médiévale obtenues par LA-ICP-MS ; le nombre de décimales est liée à la précision des mesures

Sample

Na2O

MgO

Al2O3

SiO2

K2O

CaO

TiO2

Cr2O3

MnO

Fe2O3

CuO

PbO

Gl-P-5

1.72

1.83

12.1

69.3

3.42

5.09

0.55

0.11

0.16

6.65

0.02

<0.001

Gl-P-6

1.41

2.46

10.2

73.5

2.12

5.33

0.59

0.03

0.12

5.14

0.01

0.53

Gl-P-7

1.21

2.41

13.4

72.5

0.98

4.85

0.69

0.02

0.21

3.89

0.01

0.67

Gl-P-8

1.12

3.02

15.9

70.0

1.40

2.48

1.10

0.04

0.28

5.61

0.04

<0.001

Gl-P-9

0.82

1.23

16.9

72.7

2.20

4.15

0.86

0.01

0.02

1.64

0.02

1.54

Gl-P-10

0.67

0.95

13.8

68.0

2.77

11.6

0.5

0.01

0.02

1.93

0.01

1.05

Gl-P-11

0.64

1.1

7.75

64.6

2.95

19.6

0.97

0.01

0.02

2.44

0.01

<0.001

Gl-P-12

1.12

2.74

15.9

67.7

4.41

2.3

1.06

1.02

0.14

5.77

0.01

0.04

Gl-P-13

1.73

2.7

12.9

73.0

1.15

4.88

0.67

0.01

0.50

2.82

0.01

0.42

Gl-P-14

1.62

1.9

8.54

74.6

2.24

4.27

0.62

0.01

0.09

6.16

0.01

<0.001

Gl-P-15

1.78

2.17

11.5

68.2

2.78

2.40

1.46

0.02

0.12

9.54

0.02

0.31

25The data show that the ceramic pastes consist mainly of SiO2 (64-75 %), Al2O3 (9-17%), CaO (1.3-18%), Fe2O3 (1.6-10%), K2O (1-4%), MgO (1-3%) and Na2O (0.7-1.8%).

26The results from the chemical analyses of the glazes (sample Nrs. from 5 to 15) are summarized in Table 4. Figure 3 presents the ternary diagram – SiO2, PbO, and the sum of concentrations of Na2O+K2O+MgO+CaO+Al2O3. The average metal concentrations are about 46% PbO, 39% SiO2, 7% Al2O3, 2% Fe2O3, 1% K2O, 1% MgO and Na2O<1%. The concentration of CaO is about 2%, with the exception of Gl-P-10 and Gl-P-11, glazed over white paste, both of them dated to 10th-12th century AD and imported from Byzantium. The investigated samples could be considered as high-lead glazes although the concentrations of PbO are lower compared to the glaze from other sources (see e.g. Tite et al., 1998). Literature data confirm the use of lead glazes on the Balkans as early as the 7th century AD (Walton and Tite, 2010).

Figure 3: Ternary diagram of the concentration of SiO2, PbO and the sum of Na2O+K2O+MgO+CaO+Al2O3. / Diagramme ternaire de la concentration de SiO2, PbO et la somme Na2O + K2O + MgO + CaO + Al2O3

Figure 3: Ternary diagram of the concentration         of SiO2, PbO and the sum of Na2O+K2O+MgO+CaO+Al2O3. / Diagramme ternaire de la         concentration de SiO2, PbO et la somme Na2O + K2O + MgO + CaO + Al2O3

27As discussed by Tite et al. (1998), two primary methods for production of lead glazes: either lead oxide (or some other lead compound) is applied to the surface of the pottery body, or a mixture of lead oxide and quartz. The two primary glazing methods can be distinguished by subtracting the percentage of lead oxide and any intentionally added oxide as a colorant (e.g. copper oxide) from the glaze composition and renormalizing the resulting composition to 100% (Hurst and Freestone, 1996; Walton and Tite, 2010). The concentration of the metal oxides in the glaze composition after subtraction of percentages of PbO and CuO is presented in Table 5.

4. Discussions

Chemical composition of the body

28Table 3 presents the results from the LA-ICP-MS analysis of the clay pellets. Based on the concentration of CaO clay bodies are classified in two groups. Figure 4 presents the distribution of the investigated samples (bodies and glazes) based on the concentration of CaO and Al2O3. Most of the samples dated to 10th-18th century AD have non-calcareous clay body (CaO≤5%). The samples from this group have red color due to high iron oxide concentration (3-10%), low temperature and oxidizing atmosphere of firing. Under these conditions iron oxide, present at relatively high levels predominantly occurs in the form of hematite, which contributes to the red hue of the pottery (Molera et al., 1997).

Figure 4: Distribution of the samples (bodies and glazes) according to CaO and Al2O3 concentrations / Distribution des échantillons (corps et glaçures) selon les concentrations de CaO et d'Al2O3

Figure 4: Distribution of the samples (bodies           and glazes) according to CaO and Al2O3 concentrations / Distribution           des échantillons (corps et glaçures) selon les concentrations de CaO           et d'Al2O3

29Comparison with other samples from Bulgaria is rather difficult, because of the limited number of analytical data. However, non-calcareous medieval unglazed and glazed ceramic artifacts were also discovered in other places, e.g. during archaeological excavations at the monastery of Karaachteke near Varna (Bulgaria). Both the unglazed and glazed artifacts contained a certain amount of coloring oxides (Fe2O3 + TiO2), which determined their slightly red color (Yoleva et al., 2015).

30Two of the samples from this group show different chemical composition and have grey (Gl-P-8, dated to 11th century AD) and white (Gl-P-9, dated to 13th-14th century AD) colored bodies.

31Sample Gl-P-8 represents a handle and is entirely glazed. Despite the presence of high iron oxide concentration, the body of this sample has grey color. This result may be ascribed to firing in a reducing atmosphere at temperature over 850°C (temperature reached while reduction is performed) and the presence of well-developed hercynite and Fe2+ in the silicate phase (Molera et al., 1997). The ceramic body is brown glazed which is related to the presence of Fe3+ in oxidizing environment. The reason for this effect is that reducing atmosphere is created during the firing process by the gases produced inside the ceramic body, enclosed by glaze on all sides. During firing, reducing and inert gases such as CO, water vapor, CO2, etc. are produced which consume oxygen inside the kiln and the pottery cannot be completely oxidized. During the cooling phase, there is no burning, and the partial pressure of oxygen is 20%, and the iron of the paste is oxidized while the temperature is relatively high (over 500°C). If all sides are glazed the ceramic body is sealed before the oxidizing atmosphere reaches the paste and, thus, it remains reduced due to the protective action of the glaze (Molera et al., 1997).

32Sample Gl-P-9 is characterized by lower Na2O, MgO and Fe2O3 and higher PbO concentrations. Since the sample is glazed on both sides, the white color of the body can be also related to a reducing atmosphere during the firing process. It is also possible that the high CaO content (three times higher than Fe2O3) may contribute to the white color of the body.

33The results show that two of the samples (Gl-P-10 and Gl-P-11) which are dated from 10th to 12th century AD and imported to medieval Bulgaria from Byzantium have white colored calcareous clay bodies. They are characterized by high CaO (>10%) and low MgO (~1%), Na2O (>1%) and Fe2O3 (~2%) concentrations. Besides the high CaO concentration, sample Gl-P-10 is also glazed on both sides, which may affect the color of the body.

34Some of the ceramic bodies were investigated by SEM-EDX. The selected samples were representative of the three described groups: non-calcareous – red color, non-calcareous – grey color and calcareous white color. The obtained images are presented in Fig. 5 to Fig. 7 and demostrate the effect of glazing in reducing and oxidizing conditions on the property of the clay bodies. Fig. 5 presents the SEM image of a red, non-calcareous sample (Gl-P-7), showing vitrification of the ceramic body. The next image (Fig.6) shows the fine nature of the slip layer of the non-calcareous, grey sample (Gl-P-8). The different structure of the body of sample Gl-P-8 compared to the “one-side” glazed non-calcareous samples is explained with the reducing conditions during the firing process. According to Rekecki et al., 2014, the Fe2+/Fetotal ratio played a decisive role in the sintering process under a reducing atmosphere. The formed glassy phase, with the presence of Fe2+, attacked the earth-alkali oxides after the carbonate decomposition and more effectively filled smaller pores, resulting in changed microstructural characteristics.

Figure 5: SEM image of the body and part of the glazed layer of sample Gl-P-7 / Image SEM du corps et une partie de la couche glaçurée d'échantillon Gl-P-7

Figure 5: SEM image of the body and part of           the glazed layer of sample Gl-P-7 / Image SEM du corps et une partie           de la couche glaçurée d'échantillon Gl-P-7

Figure 6 : SEM image of the body of sample Gl-P-8. / Image SEM du corps d'échantillon Gl-P-8

Figure 6 : SEM image of the body of sample           Gl-P-8. / Image SEM du corps d'échantillon           Gl-P-8

Figure 7: SEM image of the body and part of the glazed layer of sample Gl-P-10 / Image SEM du corps et une partie de la couche glaçurée d'échantillon Gl-P-10

Figure 7: SEM image of the body and part of           the glazed layer of sample Gl-P-10 / Image SEM du           corps et une partie de la couche glaçurée d'échantillon           Gl-P-10

35The SEM-image of the calcareous, white ceramic body of the sample Gl-P-10 is shown in Fig. 7. The image shows the presence of bridges between the clay body and the glaze. The dark areas in the glaze are most likely CaO inclusions.

Chemical composition of the glaze

36Metal concentrations in the glazes are presented in Table 4. In order to check the effect of the ceramic body in the quantification of the glazed layer, calculations for a layered structure were performed.

Table 4: Chemical composition of the glazes (normalized to 100%) from medieval Bulgaria using PIXE-PIGE; the number of the decimal places is related to the precision of the measurements / Tableau 4 : Composition chimique des glaçures (normalisée à 100%) de la Bulgarie médiévale obtenue par PIXE-PIGE ; le nombre de décimales est liée à la précision des mesures

Table 4: Chemical composition of the glazes           (normalized to 100%) from medieval Bulgaria using PIXE-PIGE; the           number of the decimal places is related to the precision of the           measurements / Tableau 4 : Composition chimique           des glaçures (normalisée à 100%) de la Bulgarie médiévale obtenue           par PIXE-PIGE ; le nombre de décimales est liée à la précision des           mesures

37The calculations assumed that the glaze layers were infinitely thick. For checking the influence of this assumption on the deduced elemental concentrations we performed model calculations for a target composed of a layer of PbO (with a density of 9.35 g/cm3) sitting on an infinitely thick substrate. In the first step of calculation we generated a ratio of X-rays yields for this structure, and in the second step we calculated the concentrations from these yields for an infinitely thick homogeneous target. First we assumed a layer of PbO on a substrate of SiO2. The contribution of Si K rays was generally quite low on account of lower proton energy and strong X-ray attenuation in the PbO layer. A significant apparent concentration of Si in PbO (higher than 1%) was obtained for PbO thicknesses smaller than 3 μm. We further performed calculation for more penetrative X-rays, selecting a substrate of Fe2O3. Apparent concentration of Fe in PbO smaller than 1% was obtained for PbO thicknesses smaller than 10 μm. However, as the concentration of iron oxide in the pottery body is about an order of magnitude smaller than in pure Fe2O3 (Table 3), we also determined the thickness of PbO which yields apparent concentration of iron below 10%; it was 6 μm. The range of critical PbO thicknesses is thus smaller than our actual values in the glaze, so the effect of final glaze thickness may be neglected.

38Furthermore, the obtained SEM images showed that the actual thickness of the glaze is more than 20 μm, which ensures that the quantitative results are not influenced by signals that could otherwise emanate from the ceramic body.

39The data show that no significant difference may be observed among the metal oxide concentrations depending on the sample dating and type of clay used. The analytical data indicate that these glazes are of the high-lead type. Compared to the eutectic composition, the glaze contains more silica and lower lead oxide, which suggests firing at higher temperature and high viscosity of the mixture, which would hampered the formation of thin layer.

40Higher silicon and lower lead oxide concentrations is also visible from the plotted values in Fig. 3 compared to the diagram given by Tite et al. (1998). Similar tendency is observed when comparing the results to literature data for lead glazed potteries from France, England, Serbia and Romania (Walton and Tite, 2010) and Italy (De Benedetto et al., 2004; Walton and Tite, 2010). On the contrary, the present results show lower silicon and higher lead oxide compared to lead glazes composition from Paphos region, Cypros (Charalambous et al., 2010) and similar silicon and lower lead compared to lead-glazed pottery from Paterna, Spain (Molera et al., 1997). This result suggests local production of the studied samples from medieval Bulgaria. The relatively low Na2O concentration is indicative that the glaze was not prepared according to the Islamic recipe which is characterized with high concentration of sodium (see e.g. Kuleff, 2012, 350).

41The presence of K, Al, Ca and Fe oxides in the glaze may be partially attributed to the composition of the raw materials used to make the glaze (Molera et al., 1997), but also to the diffusion of the components from the paste to the glaze during the firing. An evidence for the diffusion of CaO from the body to the glaze is the high concentration of CaO in the glaze of samples Gl-P-10 and Gl-P-11 (see Fig. 4). The rest of the metal oxides have similar concentrations in the calcareous and non-calcareous samples which leads to the assumption of diffusion processes rather than using another source. The diffusion from the clay to the glaze of K, Al, Fe and Ca has also been assumed by Molera et al. (1997). According to Walton and Tite (2010) when the re-normalized lime content (as described in the results section, see Table 4) in the glaze is higher than in the body, this is probably due to efflorescence to the surface of lime-rich salts in the clay and/or in the hard water used in forming the body. When, the iron oxide content in the glaze is higher than in the body, this is probably due to its addition to the glaze as a colorant. The composition of the present samples shows that CaO content is higher in the body clay which excludes the efflorescence process. According to the described criteria iron oxide was intentionally added as a colorant in samples Gl-P-10 and Gl-P-11 imported from Byzantium and Gl-P-12-1 from medieval Bulgaria.

42The colors of studied samples are yellow, green, brown, and black. As can be seen from Figure 2 most of the samples are decorated with more than one color which proves that both iron and copper oxides are intentionally added as colorants. The yellow color of the studied glazes is due to Fe3+ (~1%), present most probably as hematite in oxidizing atmosphere.

43Elemental analysis of the studied glazes shows that the green color is due to the presence of copper oxide. Although the studied green glazes contain iron oxide at concentrations about 1% its presence is not responsible for the glaze color. Green color is caused by the presence of Fe2+ under reducing conditions. The green glazed ceramics were certainly fired in oxidizing conditions, confirmed by the red body clay and the combination of green-yellow, green-brown and green-yellow-brown glazes in one sample, which is an indication that copper is in the form of Cu2+, and determines the green color of the glazes (Davis and Stocker, 2013). The brown color is due to the presence of high concentration of Fe3+ (~2%), or to the combination of Fe3+ and Mn3+ (samples Gl-P-8, Gl-P-11 and Gl-P-13-2). The black color of the sample Gl-P-12-1 is caused by the intentional addition of Fe3+ at concentration around 10%.

44Sample Gl-P-5 deserves more attention. Although the ceramic body is red colored, which is due to the presence of Fe3+ in oxidizing atmosphere, having in mind the low concentration of copper oxide (0.21%) in the green glaze, its color is more likely due to high iron oxide concentration (1.43%) using reducing condition. The yellowish colored glaze seems to be a second layer, covering the green glaze, applied subsequently in oxidizing environments. The yellowish color could be explained with the presence of Fe3+ in oxidizing atmosphere. Due to environmental erosive factors, the yellowish layer is partially removed (see Fig. 2).

Method for glaze production

45Table 5 shows the composition of the glazes (sample Nrs 5-15) after subtraction of percentages of PbO and CuO and re-normalized to 100%. It has been pointed out that when silica and alumina data points fall on the unity line, the glaze and body composition are essentially the same and, therefore, glazing was by the application of lead oxide by itself. In contrast, when the silica contents are richer in the glaze than in the bodies, and the alumina and lime contents are richer in the bodies than in the glaze, glazing was by the addition of a lead-plus-quartz mixture (Walton and Tite, 2010). In a third variant, proposed by Tite et al., 1998 lead compound and silica are first fired together to produce a frit which was then ground to a powder before being used to prepare the suspension.

Table 5: Chemical composition of the glazes – renormalized to 100% after subtraction of percentages of PbO and CuO / Tableau 5 : Composition chimique des glaçures renormalisée à 100% après la soustraction de PbO (en pourcentage) et CuO (en pourcentage)

Table 5: Chemical composition of the glazes –           renormalized to 100% after subtraction of percentages of PbO and           CuO / Tableau 5 : Composition chimique des glaçures renormalisée à           100% après la soustraction de PbO (en pourcentage) et CuO (en           pourcentage)

46In order to distinguish between the two primary methods of glazing, the adjusted glaze composition of SiO2, CaO, Al2O3 and Fe2O3 were plotted versus the corresponding body composition and are presented in Fig. 8.

47The plots, presented in Fig. 8 show that while for lime (R2=0.92) and alumina (R2=0.87) proportionality between glaze and clay body content is established, for silica the data are poorly correlated (R2=0.67).

Figure 8: Two-dimensional diagrams of the renormalized glaze composition of SiO2, CaO, Al2O3 and Fe2O3 vs. the corresponding body composition / Diagrammes deux dimensionnels des compositions renormalisées de SiO2, CaO, Al2O3 et Fe2O3 de la glaçure et du corps correspondant

Figure 8: Two-dimensional diagrams of the           renormalized glaze composition of SiO2, CaO,           Al2O3 and Fe2O3 vs. the corresponding body           composition / Diagrammes deux dimensionnels des           compositions renormalisées de SiO2, CaO, Al2O3 et Fe2O3 de la           glaçure et du corps correspondant

48The concentration of SiO2 and Al2O3 are equal (Gl-P-8) or higher in the body of both calcareous clay bodies (Gl-P-10 and Gl-P-11). The higher lime contents observed in the glazes of these samples could be due to efflorescence to the surface (Walton and Tite, 2010). Therefore, the glazing of this group of samples was performed by the application of lead oxide by itself.

49In most of the non-calcareous samples the concentration of silica is higher in the glaze than in the body composition, which could be considered as evidence for glazing by the addition of a lead-plus-quartz mixture. However, this would suggest lower alumina and lime content in the glaze than in the body (Walton and Tite, 2010). The close concentrations of alumina and calcium in the body and the glaze in some of the samples (see e.g. Gl-P-6.1, Gl-P-7-2, Gl-P-9; Gl-P-12.2, Gl-P-13.2) contradict this assumption.

50It might be also assumed that silica was added to the lead compound to prepare the suspension (Tite et al., 1998).

51Though, the presence of samples, with different colored glazes, in which the concentration of silica is higher (Gl-P-7.1) and comparable (Gl-P-7.2) or lower (Gl-P-12.1) and higher (Gl-P-12.2) than the corresponding concentration in the clay bodies further confirms that neither quartz nor silica were added to the glaze. Thus, it can be concluded that both the calcareous and the non-calcareous samples were glazed using lead oxide by itself.

52The plot, presenting the adjusted glaze composition of Fe2O3 versus the corresponding body composition showed rather scattered distribution of the respective points (R2=0.22) which may be due to the intentional addition of iron oxide as coloring agent.

53Because of the limited number of samples analyzed in the present work and according only to the elemental composition, a precise geographic location of the raw material sources is not possible to define. However, it should be pointed out that during archeological excavations medieval kilns for pottery production were discovered in Drastar (Angelova and Koleva, 2005; Koleva, 2008) which is evidence for the local production of the studied glazed potteries. Production of lead-glazed pottery using non-calcareous clay in combination with lead oxide by itself has been established by Walton and Tite (2010) for samples produced in the Balkans (Serbia and Romania) from the 1st to the 4th centuries AD. Futhermore, lead-isotope analysis of lead-glazed pottery found in Eastern Europe, proved that the lead for the glaze was obtained using ores from Bulgaria (i.e. the Rhodopi mountains) and it was concluded that the pottery was produced locally within the Danubian region (Iliev, 2006; Walton and Tite, 2010).

5. Conclusion

54The study of 15 medieval lead-glazed ceramics from Drastar, Bulgaria provides significant information on the chemical composition and the techniques used for their production. In particular the present study has shown that both calcareous and non-calcareous clay bodies were used for pottery production. The calcareous clay bodies have white color due to the high lime content. The majority of the non-calcareous clay bodies are red colored, determined by the high iron oxide concentration in the form of hematite. Despite the presence of high concentration of iron oxide, two of the samples from this group have grey (Gl-P-8) and white (Gl-P-9) colored clay bodies. Both of them are entirely glazed which creates a reducing atmosphere inside the clay during the firing process and the presence of hercynite and Fe2+.

55Concerning the glazes, the yellow color is due to the presence of Fe3+, the brown to Fe3+ in higher concentration or to the combination of Fe3+ and Mn3+ and the green color to Cu2+.

56Further studies on more samples excavated in Bulgaria are needed to extend the knowledge of glaze pottery production in Bulgaria.

Haut de page

Document annexe

  • Tableaux en ods (application/vnd.oasis.opendocument.spreadsheet – 17k)
Haut de page

Table des illustrations

Titre Figure 1: Map of Bulgaria with the location of Silistra / Carte de la Bulgarie avec l'emplacement géographique de Silistra
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-1.jpg
Fichier image/jpeg, 263k
Titre Figure 2: Pictures of some of the glazed pottery samples / Photos de certains des échantillons de poterie glaçurée
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-2.jpg
Fichier image/jpeg, 1,1M
Titre Figure 3: Ternary diagram of the concentration of SiO2, PbO and the sum of Na2O+K2O+MgO+CaO+Al2O3. / Diagramme ternaire de la concentration de SiO2, PbO et la somme Na2O + K2O + MgO + CaO + Al2O3
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-3.jpg
Fichier image/jpeg, 171k
Titre Figure 4: Distribution of the samples (bodies and glazes) according to CaO and Al2O3 concentrations / Distribution des échantillons (corps et glaçures) selon les concentrations de CaO et d'Al2O3
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-4.jpg
Fichier image/jpeg, 218k
Titre Figure 5: SEM image of the body and part of the glazed layer of sample Gl-P-7 / Image SEM du corps et une partie de la couche glaçurée d'échantillon Gl-P-7
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-5.jpg
Fichier image/jpeg, 275k
Titre Figure 6 : SEM image of the body of sample Gl-P-8. / Image SEM du corps d'échantillon Gl-P-8
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-6.jpg
Fichier image/jpeg, 428k
Titre Figure 7: SEM image of the body and part of the glazed layer of sample Gl-P-10 / Image SEM du corps et une partie de la couche glaçurée d'échantillon Gl-P-10
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-7.jpg
Fichier image/jpeg, 279k
Titre Table 4: Chemical composition of the glazes (normalized to 100%) from medieval Bulgaria using PIXE-PIGE; the number of the decimal places is related to the precision of the measurements / Tableau 4 : Composition chimique des glaçures (normalisée à 100%) de la Bulgarie médiévale obtenue par PIXE-PIGE ; le nombre de décimales est liée à la précision des mesures
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-8.jpg
Fichier image/jpeg, 619k
Titre Table 5: Chemical composition of the glazes – renormalized to 100% after subtraction of percentages of PbO and CuO / Tableau 5 : Composition chimique des glaçures renormalisée à 100% après la soustraction de PbO (en pourcentage) et CuO (en pourcentage)
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-9.jpg
Fichier image/jpeg, 500k
Titre Figure 8: Two-dimensional diagrams of the renormalized glaze composition of SiO2, CaO, Al2O3 and Fe2O3 vs. the corresponding body composition / Diagrammes deux dimensionnels des compositions renormalisées de SiO2, CaO, Al2O3 et Fe2O3 de la glaçure et du corps correspondant
URL http://journals.openedition.org/archeosciences/docannexe/image/4894/img-10.jpg
Fichier image/jpeg, 209k
Haut de page

Pour citer cet article

Référence papier

Valentina Lyubomirova, Žiga Šmit, Helena Fajfar et Ivelin Kuleff, « Determination of the chemical composition of medieval glazed pottery from Drastar (Bulgaria) using PIXE/PIGE and LA-ICP-MS »ArcheoSciences, 41-1 | 2017, 69-82.

Référence électronique

Valentina Lyubomirova, Žiga Šmit, Helena Fajfar et Ivelin Kuleff, « Determination of the chemical composition of medieval glazed pottery from Drastar (Bulgaria) using PIXE/PIGE and LA-ICP-MS »ArcheoSciences [En ligne], 41-1 | 2017, mis en ligne le 21 juin 2019, consulté le 19 janvier 2022. URL : http://journals.openedition.org/archeosciences/4894 ; DOI : https://doi.org/10.4000/archeosciences.4894

Haut de page

Auteurs

Valentina Lyubomirova

Faculty of Chemistry and Pharmacy, University of Sofia, James Bourchier 1, 1164-Sofia, Bulgaria. (vlah[@]chem.uni-sofia.bg)

Articles du même auteur

Žiga Šmit

Faculty of Mathematics and Physics, University of Ljubljana, 19, Jadranska Str., SI-1000 Ljublana, Slovenia – Jožef Stefan Institute, 39, Jamova Str., SI-1000 Ljublana, Slovenia

Helena Fajfar

Faculty of Mathematics and Physics, University of Ljubljana, 19, Jadranska Str., SI-1000 Ljublana, Slovenia

Ivelin Kuleff

Faculty of Chemistry and Pharmacy, University of Sofia, James Bourchier 1, 1164-Sofia, Bulgaria

Articles du même auteur

Haut de page

Droits d’auteur

Article L.111-1 du Code de la propriété intellectuelle.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search