3.4 Elemental XRF analyses
A series of locations of the clay substrate and green surface of the tile fragments has been the object of XRF analyses. Minor differences have been observed in the results. Consequently, the average elemental composition of the clay substrate and decorated surface are presented in Table 1. Only the most significant elements for these samples are collected in the table.
Table 1
Average XRF data for the clay substrate and decorated surface of the tile fragments.
Clay substrate | | Tile surface |
Element | % | | Element | % |
Fe | 3.911 | | Pb | 32.594 |
Ca | 2.612 | | Ca | 1.446 |
K | 2.531 | | Cu | 1.001 |
Ba | 0.051 | | Fe | 0.560 |
Co | 0.044 | | As | 0.273 |
Zr | 0.021 | | Bi | 0.095 |
Pb | 0.021 | | Sn | 0.011 |
Sr | 0.019 | | Sr | 0.006 |
Ti | 0.008 | | Zr | 0.006 |
Rb | 0.006 | | V | 0.003 |
It is evident that lead, calcium and copper metals dominate in the green decorated surface. However, iron, calcium and potassium are dominant in the clay body of the tiles. These results and those of EDX analyses indicate that lead enamel/glaze have been used in the tiles.
3.5 Raman spectra
The Raman spectra of the vitreous coating (enamel or glaze) of ceramics are very useful to estimate their polymerisation degree and the firing technology applied [24–34]. According to the methods developed for these purposes, the Raman spectra obtained from the enamel of the ceramic fragment a, Fig. 1a, have been the subject of some numerical treatments, Fig. 5. Eight points baseline correction has been applied using the spectral intensities at 163, 252, 365, 568, 702, 798, 1069 and 1192 cm− 1, Fig. 5b. The Rayleigh line gives rise to an exponential increase of intensity in the 330-0 cm− 1 region. The polymeric units of the aluminosilicate networks of the enamel can be investigated through the relative intensities of Si–O stretching and bending modes, at ~ 1000 and 500 cm− 1 respectively [24–30, 32–34]. Therefore, an appropriated spectral region to carry out this type of studies is between 350 and 1200 cm− 1, Figs. 5c and 5d.
The intensity and wavenumber of the stretching and bending modes of the Si-O bonds in this spectral region depends on the different coordination of SiO4 groups in the enamel. This results in wavenumber shifts and intensity differences of the stretching and bending modes of the Si-O bonds [24–30, 32–34]. Thus, a curve fitting procedure to the spectral profile is highly suitable to associate different SiO4 coordination groups to band components of the spectral profile. The number of components depends on the silicate coordination number Qn (Q0, Q1, Q2, Q3 and Q4) between 400 and 1100 cm− 1 [24, 27, 29, 33]. In addition, no negative component must be accepted in the final fitting. For the band at 976 cm− 1 deconvolution of the stretching components [26] are considered in the following intervals: Q0 → 700–850 cm− 1, Q1 → 800–950 cm− 1, Q2 → 900–1050 cm− 1 and Q3-Q4 → 1050–1200 cm− 1. The corresponding bending components [25] in the spectral profile around ~ 500 cm− 1 are: Q0 → 300–370 cm− 1, Q1 → 400 cm− 1, Q2 → 470 cm− 1, Q3 → 400–640 cm− 1 and Q4 → 670 cm− 1, Fig. 5d. The full width at half height (FWHH) of the band components of silicate glasses within the 250–1200 cm− 1 spectral range is usually FWHH ≥ 50 cm− 1. Consequently, this limit has been considered as the initial value of the FWHH for the Qn components to start the curve fitting process. Similarly, Gaussian components are assumed for this process considering that enamel glassy silicates are in an amorphous state [32, 33].
Silica is the main component in the manufacture of the vitreous enamels. Generally, the silica content comprises around 40–60% of the total weight of enamel. The structure of vitreous silica is produced by a three-dimensional combination of tetrahedral SiO4 units. The connectivity of these polymeric units can be investigated from the relative intensities of the components of the Si–O stretching and bending modes at ∼1000 and 500 cm− 1 respectively [28]. In highly connected structures the spectral feature around a 500 cm− 1 shows higher intensity than that of the 1000 cm− 1 region, whereas in weakly connected SiO4 tetrahedrons the opposite is observed [28]. A polymerisation index Ip has been established [26, 33] as the A500 / A1000 ratio, where A500 and A1000 are defined as the areas obtained integrating the fitted components in the spectral regions of ~ 500 cm− 1 and ~ 1000 cm− 1 respectively. The vitreous silicates can be classified according to their composition and their polymerisation index Ip [26, 28]. The resulting value of Ip for the enamel of the ceramic fragment a (Fig. 1) is 0.20, which shows that the 500 cm− 1 band area is significantly lower than the corresponding area near 1000 cm− 1, i.e. that the ν(SiO) stretching mode prevails over the δ(Si-O) bending mode in spectral intensity. In other words, the enamel silicates are formed with a low polymerisation index [24, 26, 29]. On the other hand, the wavenumber maxima υmax and δmax for the stretching and bending Si-O are 957 cm− 1 and 471 cm− 1 respectively. The value of υmax is lower than 1030 cm− 1, a the reference value found for vitreous enamels formed without modifying agents. The shift of υmax to 957 cm− 1 is typical of lead glaze/enamel [33]. A simple intensity analysis of the Raman spectral profile, Fig. 5, shows that the enamel has a low degree of chemical coordination. X-ray microanalyses confirm that Pb is the dominant element after Si. The high content of Pb gives rise to a low polymerisation index. The υmax value of the enamel when correlated with Ip is similar to those of leadcontaining or luster pottery and glasses classified as family 7 by Colomban [26, 29, 33]. The silicates with similar Si-O stretching frequencies and polymerisation index can hence be assigned to this family of enamels with a low polymerisation index. Enamels of this family contain typically between 20% and 40% percent Pb. These results are corroborated by the X-ray microanalysis of the tile, namely, a glass with a network of aluminosilicates having high Pb content. It is obvious that the introduction of Pb to the mixture leads to a change in the structural dimension of the silica network. A change of Pb content modifies the local environment of the silicate tetrahedra. The three-dimensional silica network has a tetrahedral coordination for silicon; this can be classified in agreement with the Qn coordination model, n being the number of oxygen atomic bridges for the SiO4 tetrahedral structure (n = 0–4). From the variation of the Qn distribution, it is possible to describe the silica network polymerisation status and to characterize the role of the Pb atoms. The substitution of Si by Pb does not change the Si-O matrix bonding but affects the ionic partial charge on the oxygen that modifies the Si-O length and polarizability. The ionic exchange of Pb modifies the Raman band wavenumber value to lower values than those expected [33]. This substitution of Si by Pb induces a Qn coordination degree change. There is also a relationship between the glass polymerisation index and their processing temperature Tf. The nature of the replacement cation can modify the value of Tf [26,27,29,33–35]. According to the value of the polymerisation index obtained (Ip = 0.20) for the tile fragment a (Fig. 1), the firing temperature of their enamel can be established between 600 and 800 0 C.
3.6 FTIR spectra
Infrared spectroscopy has been applied to determine the firing temperature of pottery [36–39]. The FTIR spectra of the ceramic fragments b and c have been obtained, Fig. 6, with this purpose. The bands observed at ~ 3470 and ~ 1650 cm− 1 are assigned respectively to the O-H stretching and H-O-H bending vibrations of the water molecules included in the fragments. The broad infrared absorption between 1300 and 800 cm− 1 is dominated by the Si-O stretching bands of the ceramic silicates. The maximum of this strong and broad spectral feature depends on the firing temperature used in the ceramic manufacture [38]. The absorbance maximum of the Si-O stretching region shifts towards higher frequencies and broadens with increasing temperature. This vawenumber maximum can be considered a firing temperature indicator-guide. Their value, 1075 cm− 1, and the spectral profile observed in the infrared spectra of the ceramic fragments b and c, Fig. 5, indicate that these ceramic fragments were fired at relatively low temperature, not exceeding ~ 800 °C [38]. Similar results were obtained from FTIR studies on Transylvanian pottery shards extracted from the same excavation [39].. Considering the results obtained applying different spectroscopic techniques to the three ceramic fragments a, b and c, as well as to pottery shards found in the indicated archaeological site [39], it is concluded that open fires, that reach a maximum temperature of about 800 °C [40], could have been used to process all these pieces. It is well known that lead glazes typically melt at temperatures between 700 and 1000 0C depending on the composition of the glaze [41, 42]. The results indicate that lead alkali glazes, with a high lead content, have been used to coat these fragments. A type of glazes with a content of lead that began to be used in the 10th and 11th centuries in the Western word [42, 43].