Molecular characterization of the mosaics
The different colors (white, orange, red and black) of the mosaics were analyzed in situ in order not only to obtain the mineralogical composition but also to understand the compounds used to give the different colors to each tessera. The discussion of the obtained results is grouped according to the color of the analyzed tesserae.
Red and orange tesserae
Raman measurements of the red and orange colored tesserae offered the typical spectrum of hematite as shown in Fig. 2 (Fe2O3; 227, 291, 411 and 609 cm−1 bands [1, 5]) and calcite (CaCO3; 155, 282, 712 and 1086 cm−1 bands [5]).
The use of hematite in Ancient Roman period is well known and has been widely identified in previous works. Moreover, its use in Pompeii has been clearly proven in the literature [1, 4, 5, 37,38,39,40]. Therefore, Raman results suggest that these reddish colored tesserae were obtained by using hematite red pigment layer applied over a calcite-based tessera. On the other hand, the spectra acquired by DRIFT spectroscopy, only showed the presence of calcite (see Additional file 1: Figure S2). Hematite is difficult to detect by means of DRIFTS because the main bands of red iron oxide are present in the 400–700 cm−1 spectral region. In this spectral region the energy falls down in the portable infrared spectrometer, thus it is not possible to detect any band in this wavenumber region. Thereby it was not possible to detect hematite by means of DRIFT portable spectrometer, but its presence in red and orange pictorial layers was clearly proven by means of Raman spectroscopy.
White tesserae
White mosaics, however, were composed mainly by a calcite-based matrix without the application of any pictorial layer, as shown in the obtained Raman spectrum. In Additional file 1: Figure S3 the typical bands of calcite at 282, 712 and 1086 cm−1 [5] can be observed. In this case no additional colored compounds apart from calcite were detected, suggesting that the molecular composition of these white tesserae consisted only in calcite. As observed in previous works [11, 17], in Roman white tesserae the presence of calcite-bearing rock was clearly and frequently identifiable as the mineral constituting the substrate. In this case, pictorial layers were not added to obtain different colored hues.
Apart from calcite, gypsum (CaSO4·2H2O) was also punctually identified (415, 1008 and 1136 cm−1 Raman bands [5], see Fig. 3).
Since the tesserae were subjected to a previous cleaning process, this gypsum could belong to a degradation process that is taking place in the calcite matrix of the tesserae with environmental factors such as SOx gases present in the past and current Pompeian atmosphere. These SOx gases, when oxidised, form H2SO4 aerosols with the rain water that impact mosaics, producing the degradation process of the calcite into gypsum, which it is soluble in water. This is one of the most dangerous degradations due to the fact that it might jeopardize the structure of tesserae, making possible the loss of material due to the water soluble gypsum formation. However, the presence of gypsum was only punctually detected, suggesting that this deterioration process has not affected the integrity of the matrix. Even so, the presence of this degradation product should be taken into account and some protection actions should be considered in order to protect the tesserae for their adequate preservation. Elemental data acquired by HH-EDXRF revealed also higher net counts of sulphur in specific areas of the measured white tesserae (see Additional file 1: Figure S4), reinforcing the molecular evidences about sulphates punctual presence.
DRIFTS analysis confirmed Raman results since all the acquired spectra matched perfectly with the DRIFT spectrum of calcite standard (see Fig. 4). In this case, gypsum was not detected. Considering that the presence of this sulphate is punctual and also that with the DRIFT sampling interface the measured area is higher (2 mm) than the Raman spot size (85 μm), the presence of this sulphate at minor/trace level can be diluted in the calcite matrix measured area giving rise to the non-detection of gypsum.
Black tesserae
The acquired Raman spectra of black tesserae showed the typical feature of a silicate-bearing material (see Fig. 5), most probably from local black-colored volcanic rocks.
The closeness to Mount Vesuvius completely influences the topography and the geological composition of the soils of Pompeii and surroundings. The presence of volcanic and igneous rocks creates a dark landscape, because usually the color of this type of rocks is black. In this sense, diopside (CaMgSi2O6) is a very common rock-forming mineral that gives black color to many igneous rocks. Moreover, this clinopyroxene has been recognised in the composition of the Vesuvian lavas and pyroclastic materials [41,42,43]. As can be observed in Fig. 5a, the obtained representative spectra of the black tesserae fit with the diopside standard spectrum, confirming the presence of this compound widely present in the tephritic lava from Mount Vesuvius. Apart from that, feldspar, with the most intense peaks at 481 and 512 cm−1 [44, 45] could be also present in the composition of black tesserae, as shown in Fig. 5b, in which those commented bands related to feldspar and others related to diopside can be observed. Feldspars also represent one of the main constituent of Vesuvian rocks, therefore its presence, together with the one of diopside, points out the local origin of the material used for black tesserae, as a previous work already have demonstrated [11]. In the same figure, bands at 665 and 324 cm−1 could also suggest the presence of magnetite (Fe3O4) [46].
Apart from the previously identified volcanic compounds, which were extensively detected, in some spots also leucite was detected (see Additional file 1: Figure S5). Leucite has also been identified as a ubiquitous component in the area of Mount Vesuvius [47], which once again confirms the local origin of the material used for the black tesserae.
Elemental characterization of the mosaics
As mentioned before an in situ elemental characterization was performed by means of HH-EDXRF to study possible similarities/differences on the elemental composition of the colored mosaics. For that, four to nine measurements in different tesserae of the same color were performed to have a big enough replicate measurements from each type of tesserae. Thanks to that, representative analysis and further chemometric data treatment were performed.
The multivariate analysis was performed by means of PCA using the raw net counts of the detected elements. In previous works [3, 36], a normalization of the net counts was applied to correct the obtained intensities against (a) the line of one element that showed constant levels in all the samples or (b) the Compton line. In this work, two datasets were performed to compare which one offered the best and most realistic results: one composed by the raw net counts and another dataset with the normalized net counts. As the signal of each element showed high variability depending on the color of the analyzed tessera, the normalization process gave bad PCA results with low explained variance, deciding in this sense to construct the data matrix using the raw net counts of the Kα line of each element or the Lα line in the case of Pb. The PCA obtained using the raw net counts dataset (23 samples and 17 variables) showed a real and faithful image of the elemental composition of the mosaic tesserae, with an explained variance of 72% (see Fig. 6).
As shown in Fig. 6, PC1 divides white, orange and red mosaic tesserae (placed in the negative part) from those of black color (in the positive part). The obtained PCA showed that black mosaic tesserae are the ones with the highest level in metals such as Al, Si, K, Ti, V, Fe, Cu, Zn, Rb, Sr and Zr. Most of these metals are highly present in rocks of volcanic origin, particularly from the Somma-Vesuvius area [41]. Moreover, the presence of these elements come in agreement with the compounds identified molecularly by Raman spectroscopy, such as diopside which is a silicate with magnesium (not detectable by means of HH-EDXRF) and calcium, which in the PCA is placed together with white and orange tesserae since their are composed by a calcite-based matrix. According to the literature [48], the volcanic material and tephritic lava from Mount Vesuvius is composed by K, Ti, Cu, Zn, Rb, Sr and Zr at minor/trace levels. Thus, the presence of these elements confirm the previous Raman results and demonstrate that black color was achieved using local black colored rocks present in the vicinities of Mount Vesuvius.
White colored mosaics presented higher levels of Ca, P and in a minor extent, Mn. The major element was calcium as expected, because white mosaic tesserae were manufactured by a calcite-based matrix, as shown in the results obtained by Raman spectroscopy. Thus, elemental measurements corroborate the previous results. Moreover, unlike black tesserae, white ones do not present any of the previously mentioned elements related with volcanic origin in significantly high levels. Instead of them, white tesserae together with the orange ones, presented the highest level of P and Mn among all the analyzed tesserae. However, the intensity of the Kα line of P and Mn was very low in all the cases. The case of Cl in red tesserae was exactly the same as the previous one.
As stated previously, white, orange and red tesserae are placed in the same part of the PC1. This is probably due to the common calcite-based matrix. However, there are significant differences in the PC2 axis. In our opinion, the amount of hematite (Fe mainly) explains such differences, because (a) the white tesserae did not have iron, and (b) the amount of Fe in the red tesserae is much higher than in the orange ones.
To confirm the reasons for such differences between red and orange tesserae, the mean value of Fe/Ca net counts ratio was calculated from these two tesserae (see Additional file 1: Figure S6). As shown in the figure, the orange colored samples have a lower level of Fe/Ca. The obtained mean value of Fe/Ca normalized net counts for red tesserae was 0.26 ± 0.04, while for orange ones was 0.15 ± 0.04. Thus, these results confirm that orange color was obtained by using less quantity of hematite or diluting it within the calcite matrix of the mosaic.
Finally, LIBS was used to approach the thickness of the applied pictorial layer in the case of red colored tesserae. For that, a certain red tessera was selected and subsequent pulses were performed in the same point of analysis. In order to assess the pictorial layer thickness, the levels of Ca and Fe were monitored using their characteristic lines at 393.2 and 374.55 nm respectively. As shown in Additional file 1: Figure S7, both Fe and Ca levels are maintained constant in the applied 28 pulses, suggesting that there is no change in the in-depth levels of both elements. This means that no layer change was observed in the subsequent analyses. Therefore, taking into account that in 28 LIBS pulses approximately about 140 μm were penetrated, it points out that the pictorial layer is thicker than this.