Handheld XRF and Raman equipment for the in situ investigation of Roman finds in the Villa dei Quintili (Rome, Italy)

Abstract

In the present work, a variety of fragments of frescoes coming from the Villa dei Quintili in Rome (Italy) and dating back to the II century A.D. were subjected to, first of all, an X-ray fluorescence (XRF) analysis by optimizing a portable spectrometer for non-destructive investigation in the field of cultural heritage. The innovative aspect is the ability to obtain coloured maps referring to the distribution and concentration of elements present in the sample. Other than characterization, the aim was to improve the technique for non-invasive and fast in situ analysis. It has been performed, in conjunction with portable Raman analysis, at the ruins of the Villa dei Quintili on samples of different typologies including pottery, statues and frescoes, dating back to the II/III century A.D., different in colour, support and shape. From the results, the identification of the main pigmenting agents is attempted, providing the group of archaeologists in the villa with extremely valuable information for their work too. In particular, by comparing the XRF results for frescoes analysed in the laboratory, taken from the warehouse of the villa, and XRF and Raman data of frescoes analysed in situ, for which the provenance area inside the villa was known, the context of excavation and the manufacture process for some of the former has been hypothesized.

---

Introduction

Nowadays, cultural heritage is very strictly interconnected with scientific methodologies. A modern view of archaeometry, conservation science and, generally, art analysis cannot be separated from the use of spectroscopic and imaging techniques such as, among the others, X-ray fluorescence spectroscopy (XRF),^1,2 inductively coupled plasma mass spectrometry (ICP-MS),³ scanning electron microscopy with energy dispersive X-ray spectroscopy (SEM-EDS),^2,4,5 high performance liquid chromatography (HPLC),⁶ Fourier transform infrared spectroscopy (FT-IR)^2,7 and Raman spectroscopy.⁸

Archaeometric studies deal with valuable artifacts. So, their fundamental prerequisite remains an investigation that could be as informative as possible, but, at the same time, non-invasive or, at least, micro-destructive.^9,10 In this context, in particular, the combined use of XRF and Raman spectroscopies represents a successful approach for a non-invasive study that allows for complementary compositional information, the first one at the elemental level, and the second one on a molecular scale. It has been already applied in a variety of findings including potteries,¹¹ frescoes,¹² glasses,¹³ gemstones¹⁴ and paintings.¹⁵ In fact, the elemental (qualitative and/or semi-quantitative) analysis performed by XRF becomes necessary for the unambiguous phase identification of both organic and inorganic materials by the Raman technique, and vice versa. Starting from such a characterization, a reliable hypothesis on the provenance, manufacturing technology, and state of preservation of the artefact can be made.

Recently, mobile instruments have contributed significantly to the in situ investigation of large and/or immovable objects. They have been successfully applied in the case of historical buildings,¹⁶ sculptures,¹⁷ rock art,¹⁸ and stained glass windows.¹⁹ Recently, very interesting reviews and books of the progress of mobile spectrometers in archaeometry have been published.^20,21

The present work is mainly aimed at the development of a portable instrument (ELIO spectrometer produced by GLAB s.r.l.) for a non-destructive XRF analysis in the field of cultural heritage. The innovative aspect is the ability to obtain, with the same tool used for XRF analysis, colored maps referring to the distribution and concentration of elements present in a given sample. In this sense, the first part of the study, performed in the laboratory, allowed us to test the instrumental set-up and its software. The final aim was to optimize the instrumental performances for a fast and non-invasive in situ analysis. As a case study, a variety of frescoes, potteries and statues coming from the archaeological site of the Villa dei Quintili in Rome (Italy), dating back to the II–III century A.D., have been considered, and precious information for the archaeometric characterization of the fragments has been obtained.

The Villa dei Quintili [Fig. 1], extending between the Via Appia and the Via Appia Nuova, was the largest villa of the Roman suburbs.²² In 1985 the State acquired 23 hectares of the ancient complex, also developed in neighboring properties. The site is known in the ancient cartography as Statuary (for the richness of the sculptures accumulated through the centuries) and Old Rome (for the grandeur of monumental ruins).

Fig. 1 (a) A panoramic view of the Villa dei Quintili. (b) Plan of the Villa dei Quintili and details of the sampled areas. Notes: A = representation area; B = private residences; C = basis villae; D = frigidarium; E = calidarium; F = Ludus-Viridarium; M = small bathroom; R = hippodrome; S = stadium; T = arcades.

The owners, the brothers Sixth Quintilius Condianus and Sixth Quintilius Valerio Massimo, consuls in 151 A.D., were killed by the Emperor Commodus in 182 A.D., and all their properties were confiscated, including the villa at the V mile of the Via Appia which remained an imperial property until the third century A.D.^23,24

The excavations, carried out between the end of the XVIII and the beginning of the XIX century, have yielded numerous works of art, now preserved in various museums and private collections in Italy and abroad (the Louvre, the Hermitage, the Vatican Museums, and the Torlonia Collection). Additional findings in 1925 and 1929 have been occasional: the works discovered are now exhibited in the National Museum of Palazzo Massimo and in the Villa dei Quintili itself, along with some findings of the most recent excavations.

The residential complex underwent over the centuries a sequence of different building phases until the fourth century AD, and it became a center of recycling and production of different materials, such as marble, bricks, tuffs, and metallic and glassy materials. Several materials, especially frescos fragments, were removed from their original location and reused. From the archeological point of view, the site is thus a complex case study and the analytical characterization of the pigments can support the conservators and restorers to (i) increase the knowledge on the original position of the frescos and (ii) rewrite a correct map of the different areas of the villa.

This study allowed us to identify, easily and quickly thanks to the use of portable spectrometers, the elemental constituents, by XRF, of the pigmenting agents of the analysed samples, and, as an innovative aspect, their distribution and concentration in a given sample. Nevertheless, as is well known, in the case of pigments with the same or similar colour and same main elements, or pigments with the same colour and same formula, or mixtures of pigments, any conclusion on the nature of pigments coming only from the elemental composition as obtained from XRF remains a reasonable hypothesis. So, the unambiguous identification of the pigments, starting from their elemental composition, was also attempted by a complementary inspection, by Raman spectroscopy, of their molecular nature. The obtained results are crucial in archaeological research, since the knowledge of materials and preparation methods of pigmenting agents, peculiar to a specific population and technology development, is extremely helpful for a correct historical-geographical classification of the artworks. In particular, in our case, starting from similarities observed in the elemental XRF composition of frescoes analysed in the laboratory, taken from the warehouse of the villa and hence of an unknown provenance area and for which a Raman investigation of the molecular nature of the pigments was impossible because of the organic layers that covered them, and some of the frescoes investigated in situ, of a certain provenance area and for which the Raman technique allowed the identification of pigmenting agents, a hypothesis concerning the context of excavation and the manufacturing technology of some of the former has been for the first time attempted.

The reported investigation is carried out in the framework of well-established multi-technique archaeometric research on this renowned site.^25–27 The sample selection is performed according to the NORMAL 3/80 (ref. 28) with the support of archaeologists from the Archaeological Superintendence of Rome.

Experimental

Materials

XRF laboratory analysis. XRF measurements have been performed on seven fresco fragments, dating back to the II century A.D., referred to as Cod. r19d, Cod. r19e, Cod. r19b, Cod. r19c, Cod. r19f, Cod. r63, and Cod. section. Some representative images are reported in Fig. 2 as examples. It has to be pointed out that these samples are erratic; no sample has been taken from its original position.

Fig. 2 Images of the samples analysed in the laboratory. (a) Fresco Cod. r19d, (b) Fresco Cod. r19e, (c) Fresco Cod. r19b, (d) Fresco Cod. r19c, (e) Fresco Cod. r19f, (f) Fresco Cod. r63, and (g) Fresco Cod. section.

We want to also add that since these aforementioned samples came from the warehouse of the Villa dei Quintili, they have, in the course of time, accumulated layers of organic materials on the pictorial film. So, even if Raman spectroscopy on these samples was also attempted, the presence of these layers generated a high fluorescence that completely covered the signal coming from the pigment, making the Raman results unreliable.

XRF and Raman in situ analysis. For in situ analysis we selected different samples present in the Antiquarium of the Villa dei Quintili.

Investigated samples are earthenwares, statues and frescoes; for each category, chosen samples differ in colour, shape and support.

All the information relative to the investigated samples is reported in Table 1.

Table 1 Information relative to the samples analysed in situ

Sample	Information	Performed analysis
Earthenwares
Brick with red writing	Not inventoried	XRF
Figured tile terminal	Not inventoried	XRF
Head of a man with a stucco tiara	II century A.D.	XRF
	From the area (R)
Statues
Statue of Zeus sitting on a rock	First half of the II century A.D.	XRF, Raman
	From the Villa, 1925–1926
Memorial, votive support to Zeus Brontoon dedicated by the matron Arrulia Hygia for the happy return from a trip	III century A.D.	XRF
	From the VII km of the Via Appia Nuova cd. “Oriental sanctuary”, 1929
Relief with the Phoenician goddess Astarte	II century A.D.	XRF
	From the VII km of the Via Appia Nuova cd. “Oriental sanctuary”, 1929
Alabaster bustier of Isis	II–III century A.D.	XRF
	From the VII km of the Via Appia Nuova cd. “Oriental sanctuary”, 1929
Headless statue of Heracles	II/III century A.D.	XRF
	From the VII km of the Via Appia Nuova cd. “Oriental sanctuary”, 1929
Frescoes
Fragment of a wall painting with floral elements and gems	II century A.D.	XRF
	From the Ludus Viridarium area
Fragment of a wall painting with vegetable traces	II century A.D.	XRF, Raman
	From the area between the residential sector (B) and the basis villae (C)
Fragment of a wall painting with vegetable festoon	II century A.D.	XRF, Raman
	From the area between the residential sector (B) and the basis villae (C)
Fragment of a wall painting with geometric and plant decorations	II century A.D.	XRF, Raman
	From the area between the residential sector (B) and the basis villae (C)

---

Some representative images are reported in Fig. 3, as examples.

Fig. 3 Some representative images of the samples analysed in situ. (a) Figured tile terminal, (b) head of a man with a stucco tiara, (c) memorial, votive support to Zeus Brontoon dedicated by the matron Arrulia Hygia for the happy return from a trip, (d) relief with the Phoenician goddess Astarte, (e) alabaster bustier of Isis, (f) headless statue of Heracles, and (g) fragment of a wall painting with vegetable traces.

Methods

Instrumentation. The used instrument is a portable XRF spectrometer ELIO produced by XGLAB s.r.l. The spectrometer is sensitive in the range of 1–40 keV, and it is therefore able to detect elements heavier than Na. The excitation source is an X-ray generator with a current between 5 μA and 200 μA and a voltage between 10 kV and 50 kV and with a Rh anode. The peak/background ratio is greater than 15 000. The instrument is equipped with a large area Silicon Drift X-ray Detector (SDD), with an active area of 25 mm² and a resolution energy less than 135 eV, measured at the MnK_α line (5.9 eV). As an extremely innovative aspect of this instrument, the elemental 2D mapping of the sample surface can be achieved through automatic XY raster scanning to perform elemental mapping. The translator stage can cover a maximum area of 10 × 10 cm² with an accuracy of 10 μm. With respect to other XRF mapping devices,²⁹ the compact head and the absence of any X-ray optics make it easily usable for in situ studies and ensure a good sensitivity even to lighter elements.³⁰

Raman measurements have been performed by using a portable ‘BTR 111 MiniRamTM’ (B&W TEK, USA) spectrometer with an excitation wavelength of 785 nm (diode laser), ∼280 mW maximum laser power at the excitation port, and a Charge Coupled Device (CCD) detector (Thermoelectric Cooled, TE). In addition, the laser output power can be continuously adjusted in order to obtain the best signal-to-noise ratio in the minimum integration time. The system is equipped with a fibre optic interface for convenient sampling. The excitation laser emits from the probe's end where the Raman signal is then collected from the sample. The spot size is 85 μm at a working distance of 5.90 mm.

XRF laboratory analysis. At first, an XRF punctual analysis has been performed. The measurement time was 30 s, the tube voltage was 50 kV, and the tube current was 80 μA.

After that, an ELIO spectrometer has also been used for 2D or 1D mapping of the elements. This is an innovative aspect: in this way it is possible to obtain the mapping of the elements present in a precise area of the sample. In our case, the following measurement conditions were used: measurement time per point: 1 s, tube voltage: 50 kV, tube current: 80 μA, movement delay: 100 ms, number of rows: 36, number of columns: 36, row distance: 0.5 mm, and column distance: 0.5 mm. The maps have been elaborated by using PyMCA software.

XRF and Raman in situ analysis. Punctual XRF analysis has been performed with a measurement time of 30 s, tube voltage of 50 kV, and tube current of 80 μA. XRF mapping has also been performed with a measurement time per point of 1 s, tube voltage of 50 kV, tube current of 80 μA, movement delay of 200 ms, number of rows of 30, number of columns of 30, row distance of 0.8 mm, and column distance of 0.8 mm. The maps have been elaborated by using PyMCA software.

As far as Raman measurements are concerned, the maximum power at the samples for our measurements was ∼55 mW. Spectra have been registered in the 60–3150 cm⁻¹ wavenumber range, by using an acquisition time of 40 s and a resolution of 8 cm⁻¹, adding several scans for each spectrum in order to improve the signal-to-noise ratio. An accurate wavelength calibration was performed by using an instrument from B&W Tek. Anyway, the optimal performance of the instrument was guaranteed by a calibration procedure before each measurement using the peak at 520.6 cm⁻¹ of a silicon chip. Each spectrum has been processed by subtracting the blank spectrum; in addition, a baseline correction and a smoothing process have been performed by using the BWSpec 3.27 software. The collected Raman spectra have been compared with data from databases and the literature.

Results and discussion

XRF laboratory analysis

Fresco Cod. r19d. Data have been collected on four different points: 01-d from the base_1, 02-d from the base_2, 03-d, white pigment, and 04-d, blue pigment. Fig. 4 shows the XRF spectra of the various points analysed and the magnification of some identifying areas.

Fig. 4 (a) XRF spectrum of the fresco Cod. r19d. In the inset, a photograph of the sample together with the analysed points is shown. (b) Magnification of the XRF spectrum corresponding to the Ca-K_α and Ca-K_β peaks. (c) Magnification of the XRF spectrum corresponding to the Cu-K_α and Cu-K_β peaks. 01-d: black line, 02-d: blue line, 03-d: red line, and 04-d: green line.

The points 01-d and 02-d (points of the base) show similar counts of Ca, higher than that of the point 03-d (white pigment) and of the point 04-d (blue pigment). In the point 04-d (blue pigment) Cu is observed, with higher counts than that of the point 03-d (white point). Instead, in the points of the base Cu is completely absent. In Fig. 5 we report the maps obtained by the XRF mapping analysis.

Fig. 5 XRF maps of the fresco Cod. r19d, together with a photo of the analysed part.

The mapping confirms what was observed by point analysis: calcium and copper are, in fact, complementary. Calcium is primarily present in the ground of the fresco, whereas corresponding to the blue pigment its concentration is much lower. Conversely, copper is present only in the blue pigment and completely absent in the ground. Strontium is present in a negligible amount, while the iron map might be the result of a change in the thickness of the sample. From the results, it is reasonable to hypothesize that the used blue pigment consisted mainly of copper. The ground of the fresco is instead a stone material mainly containing calcium.³¹

Fresco Cod. r19e. Two points have been analysed: a blue pigment and a point from the base. As in the previous case, the point from the base exhibits a Ca concentration higher than that of the blue pigment. Again, the blue pigment reveals the presence of Cu, whereas in the base it is completely absent. The results of the XRF mapping are in close agreement with those observed for the fresco Cod. r19d, except for the fact that silicon is present in a negligible amount and in any case corresponding to the blue pigment. These remarks allowed us to hypothesize that the used blue pigment was mainly composed of copper and to a lesser extent silicon and strontium.

Fresco Cod. r19b. The measurements have been performed on two different points: the first one, without any pigment, and the second one coloured with a green pigment. The spectra of the point with the pigment and the point without the pigment, not reported here, are virtually equal. This might indicate that the pigment is in too low a concentration to be identified or that the thickness of the paint layer is limited. Also the XRF mapping analysis does not provide enough information to identify the pigment. The identified elements (mainly calcium and significantly less iron and strontium) are typical of the stone material used as the support for the realization of the fresco.

Fresco Cod. r19c. Four different points have been analysed: 01-c, without any pigment, 02-c, red pigment, 03-c, blue pigment, and 04-c, from the base. Fig. 6 shows the various points analyzed and the magnification of some identifying areas in the XRF spectra. In addition, the entire XRF spectrum of the different analysed points is reported in Fig. 7.

Fig. 6 Magnification of the XRF spectrum of the fresco Cod. r19c corresponding to (a) the Ca-K_α and Ca-K_β peaks, (b) the Fe-K_α and Fe-K_β peaks, (c) the Cu-K_α and Cu-K_β peaks, and (d) the Pb-L_α and Pb-L_β peaks. In the inset, a photograph of the sample together with the analysed points is given. 01-c: blue line, 02-c: black line, 03-c: red line, and 04-c: green line.

Fig. 7 XRF total spectrum of the fresco Cod. r19c. 01-c: blue line, 02-c: black line, 03-c: red line, and 04-c: green line.

In all the analysed points, Ca was identified. The points 01-c (without pigment) and 02-c (red pigment) have a very similar Ca amount, the point 04-c (base) has a slightly greater Ca amount, while the point 03-c (blue pigment) has a slightly minor Ca amount. According to the Fe concentration, it is possible to classify the points in the following ascending order: 04-c (base), 03-c (blue pigment), 01-c (without pigment) and finally 02-c (red pigment). The point with traces of the blue pigment (03-c) shows the presence of Cu, which is completely absent in all other analysed points. The point analysed corresponding to the red pigment (02-c) shows the presence of Pb, observed in a much lesser extent in the point of the base (04-c) and completely absent in the other two points. The results of the XRF mapping are reported in Fig. 8.

Fig. 8 XRF maps of the fresco Cod. r19c together with a photo of the analysed part.

It is possible to observe how calcium is primarily present in the base of the fresco, and corresponding to the red pigment, in fact, its concentration decreases. Again, the red pigment reveals the presence of calcium, iron and manganese. The study of the elemental composition of the pigments is important just in order to compare the fragments of frescos with each other and to obtain information about the technical methodology of Romans to preparing frescos. Strontium is non-uniformly distributed over the whole sample and in a negligible amount compared to the other elements.³² Cu, observed in the point analysis corresponding to the trace of blue pigment, was not identified in the XRF mapping due to the reduced amount.

Fresco Cod. r19f. Two points have been analysed: the first one, without any pigment, and the second one, coloured with a green pigment. The XRF point analysis revealed for the point without any pigment a higher Ca amount and a definitely lower Fe amount with respect to the point with the green pigment. From the XRF mapping, calcium turned out to be complementary to iron, potassium and strontium. Calcium is primarily present in the base of the fresco, while corresponding to the green pigment its concentration is much lower. Conversely iron, potassium and strontium are present mainly in the green pigment; their concentration in the points in which there is no pigment is in fact limited.

Fresco Cod. r63. The XRF point analysis has been performed on two different points: the first one, without any pigment, and the second one, coloured with a red pigment. The point without any pigment exhibits higher Ca and Fe amounts, together with a lower Hg concentration, with respect to the point with the red pigment. The XRF maps revealed that calcium is present mainly at the points where the red pigment is missing, indicating a Ca-rich stone material for the base of the fresco. The presence of mercury and sulphur in the red pigment was observed. Iron has a distribution similar to that of calcium and complementary to that of mercury and sulphur.

Fresco Cod. section. Four points have been analysed: an embedded dark point, an embedded brown point, a white point, and an embedded dark point_2. From the results, the embedded dark and brown points have a similar Ca concentration. The embedded dark_2 point has a slightly lower concentration, while the white point has a much higher concentration. The white point does not present Fe, whereas the other points, based on the Fe concentration, can be classified in the following ascending order: embedded brown, embedded dark, and embedded dark_2. XRF mapping revealed that calcium is mainly present in the white base (suggesting a stone material) and not in the brown inclusions. The brown inclusions are characterized by the presence of iron, potassium and strontium.

XRF and Raman in situ analysis

Brick with red writing. XRF data have been collected on three different points: 01-t, red pigment, 02-t, without any pigment, and 03-t, more intense red pigment. Fig. 9 shows the various points analyzed and the magnification of some identifying areas in the XRF spectra. In addition, the entire XRF spectrum of the different analysed points is reported in Fig. 10.

Fig. 9 Magnification of the XRF spectrum of the brick with red writing corresponding to (a) the Ca-K_α and Ca-K_β peaks, (b) the Fe-K_α and Fe-K_β peaks, and (c) the Sr-K_α peak. In the inset, a photograph of the sample together with the analysed points is given. 01-t: blue line, 02-t: black line, and 03-t: red line.

Fig. 10 XRF total spectrum of the brick with red writing for the three analysed points. 01-t: blue line, 02-t: black line, and 03-t: red line.

From the analysis of the XRF spectra, Ca is mainly observed in the 01-t point (red pigment), in a lower quantity in the 02-t point (pigment absent), and, finally, it is not revealed in the 03-t point (more intense red pigment). Again, the points 01-t (red pigment) and 02-t (pigment absent) have a very similar amount of Fe, and the point 03-t (more intense red pigment) has a greater Fe amount than the other two points. In contrast, Sr is mainly observed, in a similar amount, in 01-t (red pigment) and 02-t points (pigment absent), and in a lower quantity in the 03-t point (more intense red pigment). These observations allowed us to affirm that the used red pigment was an earth (inorganic pigment of differently hydrated iron oxide). The support is rather a Ca-based stone material.

Figured tile terminal. The analysis has been performed on a point containing traces of a blue pigment, and another one without any pigment. Ca is present, in both points with and without a pigment, in the same amount. Fe instead is greater at the point without any pigment. Cu is present only in the point with the blue pigment, which also shows traces of Pb. The analysis of the traces of the pigment allows us to affirm that for the realization of the work a blue pigment based on copper was used. It could be azurite, since the blue of phthalocyanine, the other copper-based pigment, is ascribed, from a chronological point of view, to a period that is too recent with respect to the one in which the sample under investigation was realized. A Ca-rich stone material constitutes the base of the terminal.

Head of a man with a stucco tiara. A single point, the cheek character, was analysed. Key elements identified with the XRF technique are Ca and Fe, present in a Ca/Fe ratio of about 7 : 1.

Statue of Zeus sitting on a rock. XRF measurements have been performed on five different points: 01-s, details of the dress, right leg without any pigment, 02-s, details of the dress, right leg with a red pigment, 03-s, details of the dress, right leg without pigment_1, 04-s, right shoulder, and 05-s, right big toe toenail. In Fig. 11 we report the various points analyzed and the magnification of some identifying areas in the XRF spectra. In addition, the entire XRF spectrum of the different analysed points is reported in Fig. 12.

Fig. 11 (a) Photograph of the analysed points of the sample, together with a magnification of the XRF spectrum corresponding to (b) the Ca-K_α and Ca-K_β peaks and (c) the Fe-K_α and Fe-K_β peaks. 01-s: blue line, 02-s: black line, 03-s: red line, 04-s: green line, and 05-s: yellow line.

Fig. 12 XRF total spectrum of the statue of Zeus sitting on a rock for the five analysed points. 01-s: blue line, 02-s: black line, 03-s: red line, 04-s: green line, and 05-s: yellow line.

Based on the increasing concentration of Ca, the analyzed points can be ordered as follows: 02-s (details of the dress, right leg with a red pigment), 01-s (details of the dress, right leg without any pigment), 03-s (details of the dress, right leg without pigment_1), 05-s (right big toe toenail), and 04-s (right shoulder). Again, based on the increasing concentration of Fe, the analyzed points can be ordered as follows: 04-s (right shoulder) and 01-s (details of the dress, right leg without any pigment) have a concentration of Fe almost identical, followed by 02-s (details of the dress, right leg with a red pigment), 05-s (right big toe toenail) and 03-s (especially details of the dress, right leg without pigment_1). From a general point of view, one can note that the concentrations of calcium and iron are inversely proportional: points having a high Fe-concentration exhibit a low Ca-concentration and vice versa. The iron traces corresponding to the red pigment may suggest the use of a clay (inorganic pigment based on differently hydrated iron oxide).

Raman measurements performed on different reddish areas of the statue allowed us to clearly identify the peaks centred at ∼166 cm⁻¹, ∼311 cm⁻¹, and ∼1116 cm⁻¹, typical of calcite, as shown in Fig. 13.

Fig. 13 Raman spectrum collected on a reddish area of the Statue of Zeus sitting on a rock.

Memorial, votive support to Zeus Bronton dedicated by the matron Arrulia Hygia for the happy return from a trip. The XRF spectra on two different analysed points, one corresponding to the letter D with a red pigment and the other one, from the base, without any pigment, didn't show any relevant difference. For both points, Ca is the main element. Surprisingly, elements related to inorganic red pigments were not identified. The colour may then be obtained with a lacquer, organic dye, not detectable by the XRF technique.

Relief with the Phoenician goddess Astarte. Taking into account the two analysed points, without any pigment and with a red pigment respectively, the first one revealed a high Ca-concentration and a lower Fe-concentration with respect to the second one. This occurrence suggests that as the red pigment an earth was used (inorganic pigment mainly based on iron oxides). The support is a stone material mainly containing calcium.

Alabaster bustier of Isis. The XRF technique applied to a single point, alabaster having a reddish colour, allowed us to identify, as main elements, Ca and Fe, present in a Ca/Fe ratio of about 20 : 1.

Headless statue of Heracles. XRF spectra have been collected on a point with traces of a red pigment, and another one without visible traces of pigments. From the results, a slightly greater amount of Ca is revealed at the point without any pigment. In contrast, a greater concentration of Fe is revealed at the point with the red pigment, together with some traces of Pb. From these observations, the red pigment used is supposed to be an earth, to which, however, a small amount of red lead has been added to obtain a precise hue. Calcium is present in support of the work that is made of marble.

Fragment of a wall painting with floral elements and gems. Three different points have been analysed (see Fig. 14): 01-a, red pigment, 02-a, brown pigment, and 03-a, yellow pigment. From the results of the XRF point analysis, the points with red and brown pigments (01-a and 02-a, respectively) show a similar Ca-concentration. Conversely, the point with the yellow pigment (03-a) shows a slightly higher Ca-concentration. The points with brown and yellow pigments (02-a and 03-a, respectively) exhibit a similar Fe-concentration. The point with the red pigment (01-a), however, shows a lower Fe-concentration, the point with the red pigment (01-a) revealed the presence of Hg, completely absent in the other two points, together with a concentration of Sr slightly greater than the other two points. These occurrences support the hypothesis that the yellow pigment could be ochre, while the red pigment, because of the presence of Hg and Sr, may be vermilion. The support is once again made of a stone material containing calcium as the main element.

Fig. 14 Photograph of the analysed part of the fragment of a wall painting with floral elements and gems, together with the XRF maps of the sample.

Furthermore, Fig. 14 shows the XRF maps obtained for this sample.

Taking into account both the point XRF analysis and XRF mapping we can hypothesize that the red pigment used was vermilion. The presence of Fe in the yellow pigment allows us to hypothesize the use of yellow ocher (natural earths, yellow coloured by hydrated iron oxides, the most important of which is Fe₂O₃·H₂O). The presence of Pb corresponding to the brown pigment could be the consequence of an alteration. Minium, an inorganic red pigment having formula Pb₂O₃, in the murals and in the case of humid environments oxidizes to brown PbO₂. Today the pigment appears brown in colour, but in the past it could have been red.

Fragment of a wall painting with vegetable traces. XRF data have been collected on three different points: a green pigment, a white pigment, and a red pigment. The XRF spectra obtained from the point analysis revealed an increasing Ca-amount passing from the red pigment to the green one and then to the white one. Again, the green pigment exhibits the highest Fe concentration with respect to the other two points. The Hg concentration increases going from the white point to the red one, and then to the green one. From the XRF point analysis it is possible to state that a Ca-based pigment has been used as a white pigment (perhaps simple calcite also used for the realization of the preparatory layer). The high concentration of iron found in the green pigment suggests the use of green earth. The red pigment shows instead the presence of both Hg and Fe: this might suggest the use of vermilion mixed with a red ground. The support is a stone material mainly containing calcium. These considerations are supported also by the XRF maps obtained from these samples, not reported here.

The observations by XRF analysis are clearly supported by Raman data collected on white, green and red pigments and reported in Fig. 15(a)–(c), respectively. In particular, the white pigment turned out to be mainly composed of calcite. Going on, for the green pigment the Raman peaks centred at ∼149 cm⁻¹, 310 cm⁻¹, 550 cm⁻¹, 640 cm⁻¹, 680 cm⁻¹, 850 cm⁻¹, 1040 cm⁻¹ and 1094 cm⁻¹ can be recognized, typically attributable to green earth. Finally, the Raman spectrum of the red pigment clearly reveals the contributions at ∼268 cm⁻¹, ∼305 cm⁻¹ and ∼365 cm⁻¹, usually ascribed to vermilion.

Fig. 15 Raman spectra of (a) white, (b) green, and (c) red pigments of the fragment of a wall painting with vegetable traces.

Fragment of a wall painting with vegetable festoon. The XRF analysis has been performed on four different points (see Fig. 16): 07-a, blue pigment, 08-a, white pigment, 09-a, red pigment, and 10-a, green pigment. The punctual XRF analysis evidenced an increasing Ca-amount passing from the blue pigment (07-a), to the red pigment (09-a), to the green pigment (10-a), and then to the white pigment (10-a). Fe, almost absent in the white pigment (10-a), has a similar concentration in the blue (07-a) and green (10-a) pigments, and, finally, reaches the highest amount in the red (09-a) pigment. Cu is observed only in the blue pigment (07-a), and, finally, Hg is revealed only in the red pigment (09-a). The XRF maps of this sample are reported in Fig. 16.

Fig. 16 Photograph of the analysed part of the fragment of a wall painting with vegetable festoon, together with the XRF maps of the sample.

Combining the point XRF analysis and XRF mapping it is possible to deduce that a Ca-based pigment (perhaps simple calcite, also used in the preparatory layer) has been used as the white pigment.³³ The blue pigment, the only one presenting Cu traces, could be azurite (pigment consisting of basic copper carbonate); while the green pigment, which shows the presence of Fe, could be green earth. Finally, the presence of Pb only in the red pigment might suggest the use of vermilion to which, however, a clay (inorganic pigment containing impure anhydrous ferric oxide) was added. Calcium is the main element of the material used as the support.

Raman spectra collected on white and red pigments (Fig. 17(a) and (b) respectively) support the experimental XRF observations. On the one side, the observation of the typical peaks of calcite allows us to classify the white pigment as calcite. On the other side, the spectrum of the red pigment exhibits peaks at ∼228 cm⁻¹, and ∼302 cm⁻¹, which are attributable to vermilion and/or red ochre, and ∼425 cm⁻¹, ∼504 cm⁻¹ and ∼625 cm⁻¹ that can be ascribed to red ochre.

Fig. 17 Raman spectra of (a) white and (b) red pigments of the fragment of a wall painting with vegetable festoon.

Fragment of a wall painting with geometric and plant decorations. XRF data have been collected on five points: 11-a, the red pigment in the decoration at the bottom, 12-a, the red pigment at the base, 13-a, the red pigment in the decoration at the top, 14-a, the blue pigment, and 15-a, the white pigment. The photograph of the various points analyzed together with some identifying area of the XRF spectra is reported in Fig. 18.

Fig. 18 Magnification of the XRF spectrum of the fragment of a wall painting with geometric and plant decorations corresponding to (a) the Ca-K_α and Ca-K_β peaks, (b) the Fe-K_α and Fe-K_β peaks, (c) the Cu-K_α and Cu-K_β peaks, and (d) the Hg-L_α and Hg-L_β peaks. In the inset, a photograph of the sample together with the analysed points is given. 11-a: blue line, 12-a: black line, 13-a: red line, 14-a: green line, and 15-a: yellow line.

Ca is observed in increasing amount passing from red pigments of the decorations and blue pigments (11-a, 13-a and 14-a, respectively), to the red pigment in the base (12-a), to, finally, the white pigment (15-a). As far as the Fe concentration is concerned, it is observed to increase going from the blue pigment (14-a), the red pigment in the decoration at the bottom (11-a), the white pigment (15-a), the red pigment of the decoration at the top (13-a), and the red pigment of the base (12-a). The presence of Cu is detected in the blue pigment (14-a) and, in traces, in the white pigment (15-a). Hg was detected in the red pigments used for decorations (11-a and 13-a). Observing the results of the punctual XRF analysis it is possible to say that a Ca-based pigment (perhaps simple calcite, also used for the realization of the preparatory layer) has been used as the white pigment. The presence of Cu in the blue pigment suggests the use of azurite (inorganic pigment mainly consisting of Cu). The red support and the red decoration show two different pigments. For the red pigment of the base red clay (inorganic Fe-based pigment) was used, while for the red pigment of the decoration vermilion (more precious than the red earth) has been used. The support is a Ca-based stone material. The XRF mapping results, not reported here, confirm the results obtained by punctual XRF analysis. Furthermore, the identification of lead corresponding to the red decorations allowed us to hypothesize that the red decorations have been made using vermilion, to which a small amount of red lead was added in order to obtain a specific shade.

Once again, the use of Raman spectroscopy supported the XRF results, allowing us to unambiguously identify the use of calcite (Fig. 19(a)) for the white pigment, and vermilion (Fig. 19(b)) for the red pigment of the decoration.

Fig. 19 Raman spectra of (a) the white and (b) red pigments of the decoration of the fragment of a wall painting with geometric and plant decorations.

Now, the XRF results for the frescoes analysed in situ “wall painting with vegetable traces”, “wall painting with vegetable festoon” and “wall painting with geometric and plant decorations”, excavated from specific areas of the villa, as reported in Table 1, and for which Raman spectroscopy allowed for the unambiguous identification at the molecular level of the pigmenting agents, have been compared with the XRF data for the frescoes analysed in the laboratory, of an unknown provenance area inside the villa and for which the molecular identification of the pigments was made impossible by Raman analysis because of their covering organic layer that gave rise to a too high fluorescence. The comparison revealed remarkable similarities in the elemental composition between the fresco “wall painting with vegetable traces” and the fresco Cod. r63, and between the fresco “wall painting with vegetable festoon” and the fresco Cod. r19c. It is then reasonable to hypothesize, for frescoes Cod. r63 and r19c, a context of excavation from the area between the residential sector (B) and the basis villae (C) (as in the case of the two aforementioned frescoes analysed in situ). Again, taking into account the Raman results observed for the frescoes investigated in situ, we can make some hypotheses on the nature of pigments. In the case of fresco Cod. r63 a red pigment vermillion, HgS precisely, could have been used for the accomplishment of the work. As far as fresco Cod. r19c is concerned, red ochre (anhydrous ferric oxide impure of clays) could have been used for the realization of the red pigment, with a certain amount of red lead added in order to obtain a particular shade of colour. The trace of the blue pigment could be azurite.

Conclusions

In the present work, an innovative use of a portable spectrometer for non-invasive X-ray fluorescence (XRF) analysis of art objects, consisting in the compositional mapping of the sample's surface, other than the usual punctual analysis, is proposed. It has been applied, as the first step, in the laboratory, on a series of fresco fragments coming from the archaeological site of the Villa dei Quintili in Rome (Italy) and dating back to the II century A.D. Other than the compositional characterization at the elemental level, the research was devoted to an improvement of the performance of the instrument for a non-destructive and fast in situ analysis.³⁴

After that, the performance of the XRF portable set-up in identifying the key-elements of a variety of colouring matters has been tested in situ on samples of different nature including potteries, statues and frescoes, dating back to the II–III century A.D. The limits of this technique in mapping surfaces not perfectly flat and concentrations of the pigments were considered. The hypothesized nature of pigmenting agents is achieved by molecular characterization non-invasively performed by means of a handheld Raman spectrometer.

The relevance of this preliminary and non-invasive approach is also due to the comparison of several samples to each other, that allowed us to ascribe the right positioning of some frescos in a specific area of the villa, a historical task requested from the conservators of the archaeological site.

Acknowledgements

The authors acknowledge Dr Rita Paris, Archaeologist Director of the Soprintendenza Speciale per I Beni Archeologici di Roma, for supplying the investigated fragments of painted plasters, withdrawn from the archaeological site of the Villa dei Quintili (Rome, Italy), and for allowing the in situ analysis. We thank Dr Michele Gironda and the staff of the XGLab company.

Notes and references

1. F. Bardelli, G. Barone, V. Crupi, F. Longo, D. Majolino, P. Mazzoleni and V. Venuti, Anal. Bioanal. Chem., 2011, 339, 3147.
2. E. Aquilia, G. Barone, V. Crupi, F. Longo, D. Majolino, P. Mazzoleni and V. Venuti, J. Anal. At. Spectrom., 2011, 26, 977.
3. B. Li, J. Zhao, K. D. Collerson and A. Greig, Chin. Sci. Bull., 2003, 48, 1219.
4. G. Montana, E. Tsantini, L. Randazzo and A. Burgio, Archaeometry, 2013, 55, 591.
5. E. Basso, C. Invernizzi, M. Malagodi, M. F. La Russa, D. Bersani and P. P. Lottici, J. Raman Spectrosc., 2014, 45, 238.
6. I. Karapanagiotis, J. Theologou, A. Lakka, A. Ozoline and C. Panayiotou, Archaeometry, 2011, 53, 587.
7. D. Barilaro, G. Barone, V. Crupi, D. Majolino, P. Mazzoleni, M. Triscari and V. Venuti, Vib. Spectrosc., 2006, 42, 381.
8. G. Barone, D. Bersani, V. Crupi, F. Longo, U. Longobardo, P. P. Lottici, I. Aliatis, D. Majolino, P. Mazzoleni, S. Raneri and V. Venuti, J. Raman Spectrosc., 2014, 45, 1309.
9. D. Miriello, M. Malagodi, S. A. Ruffolo, M. F. La Russa, G. M. Crisci, A. Pezzino, R. Galluccio, D. Barca and E. Marasco, Environ. Earth Sci., 2010, 60, 829.
10. M. F. La Russa, S. A. Ruffolo, M. Malagodi, D. Barca, R. Cirrincione, A. Pezzino, G. M. Crisci and D. Miriello, Appl. Phys. A: Mater. Sci. Process., 2010, 100, 865.
11. R. Scarpelli, R. J. Clark and A. M. De Francesco, Spectrochim. Acta, Part A, 2014, 120, 60.
12. L. Appolonia, D. Vaudan, V. Chatel, M. Aceto and P. Mirti, Anal. Bioanal. Chem., 2009, 395, 2005.
13. D. Lauwers, A. Candeias, A. Coccato, J. Mirao, L. Moens and P. Vandenabeele, Spectrochim. Acta, Part A, 2016, 157, 146.
14. Z. Petrová, J. Jehlička, T. Čapoun, R. Hanus, T. Trojek and V. Goliáš, J. Raman Spectrosc., 2012, 43, 1275.
15. G. Casuccio, K. Bunker, S. Kennedy, M. Sparrow, B. Pacolay, P. Ioannidis and R. Foulke, Microsc. Microanal., 2012, 18, 958.
16. N. Prieto-Taboada, I. Ibarrondo, O. Gómez-Laserna, I. Martinez-Arkarazo, M. A. Olazabal and J. M. Madariaga, J. Hazard. Mater., 2013, 248–249, 451.
17. C. Colombo, F. Bevilacqua, L. Brambilla, C. Conti, M. Realini, J. Striova and G. Zerbi, Anal. Bioanal. Chem., 2011, 401, 757.
18. L. C. Prinsloo, A. Tournié, P. Colomban, C. Paris and S. T. Bassett, J. Archaeol. Sci., 2013, 40, 2981.
19. P. Colomban and A. Tournié, Journal of Cultural Heritage, 2007, 8, 242.
20. P. Colomban, J. Raman Spectrosc., 2012, 43, 1529.
21. D. N. Papadopoulou, G. A. Zachariadis, A. N. Anthemidis, N. C. Tsirliganis and J. A. Stratis, Talanta, 2006, 68, 1692.
22. A. Rotondi, Forma Urbis-Itinerari nascosti di Roma Antica, ANNO XVII(2), Roma, 2012.
23. R. Frontoni, Forma Urbis-Itinerari nascosti di Roma Antica, ANNO V(7/8), Roma, 2000.
24. R. Frontoni and G. Galli, Arkos, Scienza e Restauro, 2010, vol. 25, p. 66.
25. C. M. Belfiore, G. V. Fichera, M. F. La Russa, A. Pezzino, S. A. Ruffolo, G. Galli and D. Barca, Archaeometry, 2015, 57, 269.
26. V. Crupi, G. Galli, M. F. La Russa, F. Longo, G. Maisano, D. Majolino, M. Malagodi, A. Pezzino, M. Ricca, B. Rossi, S. A. Ruffolo and V. Venuti, Appl. Surf. Sci., 2015, 349, 924.
27. V. Crupi, V. Allodi, C. Bottari, F. D'Amico, G. Galli, A. Gessini, M. F. La Russa, F. Longo, D. Majolino, G. Mariotto, C. Masciovecchio, A. Pezzino, B. Rossi, S. A. Ruffolo and V. Venuti, Vib. Spectrosc., 2016, 83, 78.
28. NORMAL 3/80, Materiali lapidei: campionamento e conservazione dei campi-oni, Roma, 1980.
29. S. Mosca, T. Frizzi, M. Pontone, R. Alberti, L. Bombelli, V. Capogrosso, A. Nevin, G. Valentini and D. Comelli, Identification of pigments in different layers of illuminated manuscripts by X-ray fluorescence mapping and Raman spectroscopy, Microchem. J., 2016, 124, 775–784,  DOI:10.1016/j.microc.2015.10.038.
30. M. Alfeld, J. V. Pedroso, M. van Eikema Hommes, G. Van der Snickt and K. Janssens, et al., A mobile instrument for in situ scanning macro-XRF investigation of historical paintings, J. Anal. At. Spectrom., 2013, 28, 760,  10.1039/c3ja30341a . s.l.: Microchem. J. 124(2016) 775–784.
31. A. Zielińska, W. Dąbrowski, T. Fiutowski, B. Mindur, P. Wiącek and P. Wróbel, X-ray fluorescence imaging system for fast mapping of pigment distributions in cultural heritage paintings, J. Instrum., 2013, 8, P10011,  DOI:10.1088/1748-0221/8/10/p10011.
32. A. Deneckere, B. Vekemans, L. Van de Voorde, P. De Paepe, L. Vincze, L. Moens and P. Vandenabeele, Feasibility study of the application of micro-Raman imaging as complement to micro-XRF imaging, Appl. Phys. A, 2012, 116, 363–376,  DOI:10.1007/s00339-011-6693-5.
33. A. Deneckere, M. De Reu, M. P. J. Martens, K. De Coene, B. Vekemans, L. Vincze, P. De Maeyerd, P. Vandenabeelee and L. Moensa, The use of a multi-method approach to identify the pigments in the 12th century manuscript Liber Floridus, Spectrochim. Acta, Part A, 2011, 80, 125–132,  DOI:10.1016/j.saa.2011.03.005.
34. E. Ravaud, L. Pichon, E. Laval, V. Gonzalez, M. Eveno and T. Calligaro, Development of a versatile XRF scanner for the elemental imaging of paintworks, Appl. Phys. A: Mater. Sci. Process., 2016, 122, 1–7,  DOI:10.1007/s00339-015-9522-4.

---