Materials science and cultural heritage

Marco Martini, Anna Galli


1 Introduction

In these short review we will introduce a few applications of Materials Science to two types of human artefacts, i.e. the use of luminescence to the study of ceramics, mainly with the aim of determining their chronology, and the use of X-ray spectroscopy, XRF in particular, in the study of paintings’ pigments. Even if a huge number of other techniques are currently used, the two examples here reviewed consider one of the most diffused material all along men’ history, i.e. ceramics, and one of the nowadays most diffused and powerful technique, i.e. XRF, that in the case of paintings has recently acquired the possibility of being portable with accuracy and sensitivity comparable to the great facilities, allowing the intervention without moving the work of art.

2 Ceramics and luminescence

One of the most cited scientific applications to the Cultural Heritage is surely luminescence dating and in particular thermoluminescence. Indeed, the luminescence properties of ceramics are exploited not only as regards thermoluminescence, but also in other applications that will be shortly reviewed here.
It is rather well known that the luminescence properties of the crystals contained in the ceramic material can be fruitfully exploited in order to date them. This is mainly carried out studying the thermoluminescence (TL) of the ceramics. This technique was proposed in the sixties of the last century and some well assessed methods are now currently used to support archaeological excavations and in determining the chronological phases of historical buildings. TL has been giving useful results also in the study of clay-cores in bronze statues. It must be noted that claycores are not always made only of ceramic material and this can give high uncertainties in this application. A short review of archaeological and building dating, together with a few examples of clay-core studies will be given in sect. 2.1.
Besides TL, a further luminescence property that is increasingly exploited is Optically Stimulated Luminescence, OSL, which is based on the de-trapping of charges due to light exposure rather than heating, as is the case of TL. The physical basis of this phenomenon is rather complex and its application must be specifically carried out on each mineral extracted from the ceramic: as an example, the stimulating light for quartz inclusions should be at 470 nm, while feldspars can be stimulated both in blue and in the infrared. OSL is currently widely applied in sediment dating, but an interesting application is under study in the field of Cultural Heritage: it is the so called “surface dating”, that can in principle determine when a surface was put in the dark and then accumulate charges in the traps due to the natural radiation environment, similarly to the TL application. Surface dating will be introduced in sect. 2.2, together with an example of application.
Apart from ceramics, the luminescence properties of other materials should in principle be used to study the chronology of historic and artistic artefacts. The first material to be considered is glass whose amorphous structure however prevents a correct dating procedure in most cases. Some exceptions will be treated in sect. 2.3, where the feasibility of chronological studies on mosaic glass tesserae will be shown, when some microcrystals are present in the glass matrix.
In a further sect. 2.4, a different kind of luminescence will be exploited to contribute to the study of specific artefacts, i.e. lustre ceramics. In this case the radio-luminescence spectra demonstrate to be a kind of fingerprint in telling apart the provenance of the objects.

2.1 Luminescence dating in archaeology and in history of architecture

Thermoluminescence (TL) dating is a powerful tool in Archaeology and its reliability has been confirmed since the eighties of the XX century. TL is a phenomenon typical of many insulating materials, which emit a weak light amount when heated up to some hundreds degrees (°C), provided that they had been previously irradiated by ionizing radiation. It is a typical dosimetric property and is exploited in health physics and radioprotection, because it records the amount of energy absorbed by a material exposed to ionizing radiation. Dating by TL is a specific application of TL dosimetry in which there is a source of practically constant irradiation, the natural radioactivity of the ceramic itself and the surrounding environment.
The dosimeters are the small inclusions always present in ceramics, typically quartz and feldspar crystals, which are constantly exposed to the mentioned radiation field. Of course, micro-dosimetry is the basis for determining the amount of each type of radiation, alpha, beta and gamma, absorbed by the inclusions of different dimensions.
The precise measurements of the dose rate, which is currently reported in unit of Gray per year (Gy/y), i.e. the amount of dose absorbed by the ceramic inclusions per unit time, is a great source of uncertainty. As a matter of fact, what is currently done is the measurement of the radioactivity content of the ceramic itself and the surrounding environment, assuming that all the emitted radiation is absorbed by the ceramic, in a radiation equilibrium. The main source of uncertainty is the water content of both the ceramic and the environment, whose presence reduces the absorbed dose by the ceramic, due to the dose absorbed by the water itself. Corrections can be taken into account and the precision of the method can by rather good, specifically when the amount of water in the past can be precisely known: this is the case of very humid or very dry environments or when the ceramic is particularly compact and reduces to extremely low level the water that can be present in the ceramic.
Specific methodologies have been developed to apply TL to chronological studies in the field of archaeological excavations and buildings. We refer to a wide literature for details in these applications, specifically to the Proceeding of the series of LED Conferences. A few notes are however to be mentioned: 1) it is necessary to take samples from the ceramic to be dated, which means that the application is destructive, even if the amount to be taken can very very small. 2) To get a sufficient precision in the date, an accurate knowledge of the radiation field of the ceramic is necessary, i.e. a precise measurement of the radioactivity content of the ceramic and the environment must be carried out. 3) As mentioned above, also the water content of the ceramic and the environment should be known.
To summarize, it is necessary to take as much sample as possible, even if we are dealing with few grams, and the surrounding environment should be very well known. This is why the most precise applications of TL dating are relative to the studies of building chronologies, because it is possible to obtain large amounts of samples and the radiation environment and climate can be known with high precision. In such cases the overall precision can be better than 4−5 % of the age.
A little lower precision is typical of archaeological excavations, even if in some cases very high precision can be reached, depending on the type of ceramics, which sometimes can present very good dosimetric properties, with high reproducibility and sensitivity. An interesting case study is given by the Basilica di S. Lorenzo Maggiore in Milan, whose date of construction is uncertain, but it is known to be already built in the VI century AD. Various destructions, restorations and re-building are documented, but the complete chronology of the monument is still under study. In a wide three-year restoration project, collaborating with art historians and architects, a huge number of samples, more than one hundred, were taken and their TL dates were obtained. Their TL response was not always accurate and repeatable. In table 1 the results relative to 24 out of 42 well dated samples are reported.
It can be noted that many samples are surely older than the Basilica itself, documenting the frequent habit of re-using older bricks taken from other structures. In this case, not far from the San Lorenzo Basilica there was a huge Arena of the Imperial Age, that was abandoned in the period of the Basilica construction. Various construction phases are documented by the results reported in table 1. Samples D124 and D125 correspond to a documented big fire in that part of the building. Of course a high-temperature heating of the sample erases the TL signal starting again the accumulation of charges in the traps.
Similarly to archaeological and monument dating, also bronze statues can be indirectly dated by luminescence. The principle is always the same, but the material to be considered changes, in the sense that clay core has to be measured, in the hope that it was affected by the high temperature needed to melt the bronze and that no contamination occurred if the statue was buried in the earth. Examples of dating of bronze statues are reported: the statue of St. Peter in the Basilica Vaticana in Rome was precisely dated to the XIII century, while many attempts were made in order to date the she-wolf, symbol of the city of Rome (Lupa Capitolina, fig. 1). In this case the material was not as dosimetric as necessary and the uncertainty was rather high. However, it was possible to exclude an Etruscan attribution. This result was confirmed by radiocarbon dating of many small pieces of carbon found in the clay-core.

2.2 Luminescence dating of surfaces

An interesting new type of luminescence dating comes from the application of OSL to the surfaces of artifacts that were put in the dark in the past. In fact surfaces of stones or marbles or bricks that were exposed to sunlight for long periods had the traps in the surface layers kept empty by light stimulation. Once covered by other stones or objects, also the surfaces started to accumulate charges in the traps. A subsequent stimulation by light of appropriate wavelength gives a luminescence emission (OSL) proportional to the dose absorbed since the darkening, in a similar way like TL. Looking at bricks, the luminescence signal from the bulk is higher than the surface one, because the former is relative to the time elapsed since the heating in a kiln, while the latter to the period when the brick was in the darkness.
This advanced application has been successfully attempted on rocks, marble and stone artifacts but not yet on bricks. A recent conservation campaign at the Certosa di Pavia gave the opportunity to sample some bricks belonging to a XVII century collapsed wall, still tied to their mortars.
This was a good condition to test this technique, comparing the dating results with precise historical data. The attempt gave satisfactory results, allowing to identify bricks surely reused and to fully confirm that the edification of the perimeter wall occurred at the end of XVII century.
In fig. 2 are summarized the results relative to six samples that showed good dosimetric behavior, being the signal proportional to the absorbed dose. It must be reminded that the luminescence emission of ceramics and bricks is mainly due to quartz and feldspars: while blue light stimulates the emission of both quartz and feldspars, IR light only stimulates that of feldspars. In this study, however, a different approach, using the poly mineral (4−10 μm)-grained fraction of the material, has been chosen. The main reason of this choice, despite the disadvantage of a luminescence emission due to the superimposition of different contributions, which forces the use of different stimulation wavelengths, is the limited amount of material required, the sample needing no chemical treatments.
Looking at fig. 2 relatively low precision is evident, but the possibility of discriminating the re-used bricks from the ones contemporary to the construction of the wall is also evident. The period 1530−1650 AD comes out from the surface data, in good agreement with historical records and archaeological evidences. Two of the bricks were surely reused, their TL date indicating the late middle age for firing in kiln (PV18 and PV23 samples). A limited number of papers have been published on “surface dating”, evidencing the troubles that derive from one of the most critical items, i.e. the erasure by light of the OSL signal. In fact, a non- rapid decrease of the surface signal as a consequence of a short exposure to light results in a limit of the applicability of surface dating technique.

2.3 Dating mosaic glass tesserae

The amorphous structure of glasses evidently limits the possibility of dating them. In fact, many recombination channels are possible in a vitreous material, which give as a consequence instability in the number of the trapped charges and fading of the luminescence signal. Various studies demonstrated that the presence of micro- and nano-crystals circumvent this limitation and allow to use glass as a dosimetric material as firstly reported by Mueller and Schvoerer in 1993.
In a study of 2003 it was put in evidence that the presence of antimony enhances the TL signal of the vitreous matrix. This opacifier possibly forms new TL centres, when producing calcium antimonate crystals (CaSb2O6).
The simple addition of antimony to a modern glass sample did not produce the stable TL signal as is the case of historic glass tesserae, where calcium antimonate is present. A comparison of the TL signal obtained from an ancient mosaic tessera from Faragola and of a CaSb2O6 standard is reported in fig. 3a and fig. 3b respectively, where the reported spectra refer to spectrally resolved measurements, using a specifically developed system.
Similarly to the TL signal produced by the presence of calcium antimonate crystals, it was observed that mosaic tesserae from the vault of the Basilica of St. Peter in Rome, opacified with tin-based additives (in particular, cassiterite, tin dioxide) showed higher luminescence sensitivity and thermal stability, being therefore more suitable for dating applications.

2.4 Radioluminescence of lustred ceramics

Lustre is a decorative metal-like film applied to a white opacified surface of glazed pottery; the tradition of decoration with this technology began in the early Islamic period and has continued in the Mediterranean area until the present days. In central Italy the production of lustred majolica achieved its maximum magnificence during the XV and the XVI century. This type of pottery consists of a ceramic body, made of fired clay, covered with a decorative glaze on which a lustre layer is applied. The glaze is a glass-like material, usually 100–300 μm thick, which contains a mixture of silica, soda and lead-alkali. The lustre film (0.1–2.0 μm thick) is the external layer, characterized by the presence of copper and silver nanocrystals embedded in a glassy matrix . The lustred ceramic is than submitted to three subsequent heating processes.
In a study of 22 samples found during excavations made in central Italy, in Egypt, or provided by private collectors their luminescence properties have been studied. The TL of the ceramic body was used to obtain the date of production. The luminescence emissions under irradiation, the so-called radio-luminescence (RL), of the different parts of the lustred ceramics were studied.
The main results can be summarized as follows:
TL dating of lustred ceramics was successfully performed on the 22 samples of different age and provenance. In most cases the results confirmed most scholars’ attribution but definitely excluded the Hispano-Moresque dating of a few of them. Two samples turned out to be Fatimid, and this result is particularly important, due to the uncommon finding of Fatimid lustred ceramic The application of RL gave interesting results for the characterization of ceramic bodies, and the preliminary results obtained on lustred ceramics are encouraging, even for the identification of technological features. Characteristic RL emissions have been in fact successfully associated to the particular manufacturing technology of the lustre, being independent of the type of clay. For the emissions relative to the glaze and the lustred glaze the presence of cassiterite is confirmed, associated with the glaze as proposed in and that the metallic silver clusters are typical of the Italian gold luster.

3 Paintings and Spectroscopy

Understanding how painted works of art were realized layer by layer, requires a wide range of spectroscopic methods. The main challenge in this research topic is the wide variety of materials to be considered, which include: inorganic or organic pigments, either natural or synthetic; bio-organic materials such as wood, paper, parchment; oil, protein, sugar-based binding media, glues, synthetic or natural varnish coatings and associated by-products arising during aging. Furthermore, given the value and fragility of artifacts, analyses need to be carried out in as non-invasive a manner as possible, often directly on the object itself or, in certain cases, on small samples.
In addressing these types of demanding analytical challenges, it has been necessary to develop and use powerful and versatile analytical instruments and methodologies.
The micro-invasive approach is particularly challenging, due to the fact that artwork materials are complex mixtures intrinsically heterogeneous, composed of a wide range of compounds, both organic and inorganic, as mentioned. The studies include a wide range of size scale, which goes from the chemical identification of compounds to the mapping of trace elements, alteration or restored phases.
It is from this intrinsic complexity that comes the need for multi-analytical approaches, to overcome the limitation of individual spectroscopic methods. The systematic application of spectroscopy techniques in the conservation science dates back to the the fifties of the XX century.
In sect. 3.1 to 3.4 the attention is focused on the analysis of painting materials due to their importance in conservation science. The precise identification of pigments can help in dating and authentication studies, since each pigment has its own and well-known period of use. Furthermore, it can help in understanding the painter technique and in tracing historic trade routes. The deep knowledge of the material and of the degradation processes are crucial for the fine tuning of the restoration and conservation protocol, since not all the pigments react in the same way to the exposure to light or environmental conditions. Several analytical techniques for the identification of pigments have been in use for many years; they include microscopy, X-ray−based techniques, UV-Vis and IR spectroscopy, laser-based techniques, such as Raman and Laser-Induced breakdown spectroscopy, and spectrometry techniques coupled with gas or liquid chromatography.
We will focus on X-ray Fluorescence (XRF), the most diffused and powerful technique, whose application is still giving extremely useful results, also as a consequence of the development of new instrumentation, more and more sensitive and reliable, mainly when coupled with other spectroscopic techniques.
A short introduction of the EDXRF analysis applied to painting study, together with a few examples will be given in sect. 3.1.
In the following, we will discuss the use of spectroscopic analytical techniques with portable instruments (sects. 3.2 and 3.3), because archaeometric analyses took great advantages from this class of spectrometers and some dedicated solutions are now available. Unambiguous identification is usually obtained with a multi-analytical combination of techniques, so in sect. 3.4 an example of FORS (Fiber Optics Reflectance Spectroscopy) coupled with EDXRF will be briefly described. Working on unique and not standardized objects requires to pay attention on details and to know how to choose correct parameters and calculation algorithms to obtain reliable results. The opportunities to deal with these objects are very limited and it must be noticed that the development of very sensitive specific methodologies and instrumentations to be applied to the Cultural Heritage has been giving useful examples for their applications to other fields of research.

3.1 EDXRF (Energy Dispersive X-ray Fluorescence)

The property of all the elements of emitting characteristic X-rays when stimulated by various types of radiation is widely exploited in the analysis of materials.
Specifically in the field of Cultural Heritage, Energy Dispersive X-ray Fluorescence (EDXRF) is surely the most diffused type of study of the composition of the materials. To excite X-ray emission various kinds of radiation are used, the main relative techniques being: PIXE (particle induced X-emission), SEM-EDS (scanning electron microscope energy dispersive emission) and synchrotron radiation in the SRXRF analysis. All of them require the work of art to be taken to the measuring facility. More recently portable XRF instruments have been developed and will be treated in sect. 3.2.
In accordance to Lahanier et al., the XRF analysis could be considered the ideal procedure for analyzing Cultural Heritage materials because is non-invasive (respecting the physical integrity of the object), but also fast (to investigate the object at various points), universal (to analyze many materials), versatile (allowing to obtain average compositional information but also permitting local analysis of small areas), sensitive (allowing to detect trace element as fingerprints) and multielemental (to obtain simultaneously information on many elements in a single measurement).
The identification of almost all the elements is possible through a qualitative analysis. It must be noted that the low energy of X-ray emitted by light elements precludes their detection: as a general rule only elements with $Z > 15$ can be detected. Much more difficult is the quantitative analysis in archaeometric applications. The irregular shape or the non-homogeneous composition of the sample have generated a widespread opinion that only semiquantitative analyses are possible in XRF applications to archaeometry. In fact, this is always true for nonhomogeneous samples like, typically, metallic corroded samples or many pigments and binder mixtures. In some cases the just mentioned limitations can be overcome and quantitative analyses have been successfully obtained. Even if detailed quantitative analysis is very useful, in the field of Cultural Heritage it is not always necessary, as also qualitative information by XRF spectra increase the knowledge of artifact.
As an example Dik and colleagues used high-energy X-rays at the L beam line of the DORIS III synchrotron facility (DESY, Hamburg, Germany) to record XRF spectra from a $17.5\times 17.5$ cm2 area of the Van Gogh painting Patch of Grass. Studying the acquired data, the authors have visualized the portrait below the visible landscape: the flesh tones of the hidden head of a woman were reconstructed by combining the antimony and mercury distributions. By the use of a proton beam, PIXE spectra of the inks of the Galileo manuscripts at the LABEC laboratory in Firenze allowed to reconstruct the chronological history of Galileo studies.
For the purpose of examining artists’ materials, as this method detects only the elements, not their chemical composition, some pigment can only be inferred. Typical examples are green pigments such as malachite and verdigris that have in common the copper, detected by XRF, but different combination of light element, as H, C, O, whose detection and quantification is not possible, as already noticed . For some pigments, such as zinc white (ZnO) or lead white ((PbCO3)2Pb(OH)2), pigment identification can be a straightforward task. On the contrary the discrimination between type I and type II of lead-tin yellow cannot be done only on the basis of their elemental composition: the Pb/Sn ratio is not always specific to the actual type of this kind of pigments. Other compounds may have different crystallographic phases, e.g. titanium oxide, which can appear as either anatase (produced since 1920) or rutile (since 1938).

3.2 p-XRF (Portable Energy Dispersive X-ray Fluorescence)

Recent advances in X-ray tube and detector technologies have allowed the development of portable X-ray Fluorescence devices (p-XRF). The combination of conveniently portable, high-resolution instrumentation capable of non-destructive, multielement analysis has underpinned the rapid growth in the application of p-XRF in the field of Cultural Heritage. This is because portable instrumentation capable of non-invasive elemental characterization has obvious benefits in the access to museum collections and archaeological sites.
However, the advantages of p-XRF are offset by limitations of the instrumentation itself (for instance lower sensitivity than lab-instruments) and analytical constraints already discussed in the previous paragraph due to the nature of the sample surface and matrix.
Like in the other kinds of XRF also in p-XRF, one must consider that the thickness of the investigated layer depends on the primary X-ray beam energy, on the analyzed elements and on the sample (matrix) mean composition. When the analyzed materials are inhomogeneous in depth, the problem of discriminating the composition of the surface layers from that of the underlying layers arises. This is the typical case of layer structured Renaissance paintings on canvas or wood: one or more paint layers are laid on a properly prepared substrate (typically calcium sulphate, plaster, or calcium carbonate, sometimes covered with a lead white or colored layer, the so-called priming or imprimatur).
To solve the problem of discriminating the composition of the painted surface layers from that of the underlying ones, in 2007 Bonizzoni et al. suggested an experimental method, that, exploiting the fundamental parameter method (FPM), allows to recognize a multilayer structure assigning to each layer the proper elements detected and the proper thickness. This assignment, cannot be given by XRF alone, being impossible to understand which layer the signals come from: for this reason a tool to recognize the composition of the external layer (surface pigments) is needed, like for instance visible reflectance spectroscopy (vis-RS).
If, as usually occurs, the sample is a sequence of layers, the characteristic line intensity Ii of element i from a layer depth t excited by monochromatic exciting radiation is detected as:

$I_{i}= \frac{I_{0} exp (- \sum\limits_{k>x} \mu_{k}^{M} (E_{0}) \rho_{k} t_{k}+ \sum\limits_{k>x} \mu_{k}^{M} (E_{i}) \rho_{k} t_{k})P_{i} \epsilon_{i} Gw_{i}\mu_{i}^{ph} (E_{0})}{\frac{\mu_{x}^{M}(E_{0})}{sin \varphi_{1}} + \frac{\mu_{x}^{M}(E_{i})}{sin \varphi_{2}}}[ 1 - exp (- (\frac{\mu_{x}^{M}(E_{0})}{sin \varphi_{1}} + \frac{\mu_{x}^{M}(E_{i})}{sin \varphi_{2}}) \rho_{x} t_{x})]$

where:
$I_{0}$ = intensity of incident monochromatic radiation,
$E_{0}$ = energy of incident monochromatic radiation,
$P_{i}$ = atomic factors,
$w_{i}$ = relative content of element i in the sample,
$m_{i}^{0}(E_{0})$ = photoelectric absorption coefficient of element i at energy $E_{0}$,
$\mu_{x}^{M}(E_{0})$ = mass absorption coefficient of matrix of x layer at energy $E_{0}$,
$\mu_{x}^{M}(E_{i})$ = mass absorption coefficient of matrix of x layer at energy $E_{0i}$,
$\varphi_{1}$ = incident angle,
$\varphi_{2}$ = exit angle, $\rho$ = density of layer,
$k$ = sequence of the layers immediately over the x layer characterized by element i.
If the element i of eq. (1) characterizes the pigment of the surface layer, which can be identified by reflectance spectroscopy measurements, the only unknown value is the layer depth, as $w_{i}$ is known by literature. Absorption coefficients are known and the geometry factors can be obtained measuring a few standards under the same conditions as the samples. As regards the density $\rho$ and the composition of matrix of the layer, a mixture of pure pigment and binder (linoleic oil) in different proportions for the pigment drawing up has been assumed in the calculation.
A large set of p-XRF (fig. 4) were performed on a famous wood painting by Andrea Mantegna - “Madonna col Bambino e un Coro di Cherubini”, Pinacoteca di Brera, Milan, in the occasion of its restoration directly in the museum room. One of the main results was the determination of the stratigraphic sequence of the Virgin’s cloak, where the azurite is covered by a velatura of lapis lazuli, as frequently found in Mantegna, and a lead white priming under the Madonna and child is present. As an example for the Virgin’s blue mantle the results obtained from the areas tagged 8, 9, 14 and 15 suggest two possibilities for painting sequence: one consists of a lapis lazuli layer over azurite mixed with lead white (2 layers); the other consists of a lapis lazuli layer over azurite laid on lead white (3 layers). The reflectance spectroscopy in the visible (vis-RS) measurements acquired where lapis lazuli layer was lost because of conservation problems, confirms that the most probable sequence is the second one, with almost no evidence of lead white in the azurite layer.
The portable scanning system developed at LABEC gave results comparable with sophisticated (SR-based) methods as described in the first example of sect. 3.1. In fact, taking advantage of the elemental imaging, it has been possible to determine not only the general composition of the painting, but also to suggest a sequence of paint layers and to find “hidden” features such as, for example, pentimenti (underlying paintings in the artwork, evidence of revision by the artist). Thanks to the collaboration of the authors with the “Opificio delle Pietre Dure” (OPD) in Florence, primary results from measurements performed with the XRF scanner on “The portrait of a young woman” by Raffaello Sanzio have been achieved. An example of the efficacy and versatility of the new XRF system is in the revealed hidden bow in the shoulder of the lady’s dress: the acquired elemental maps of Pb M series, Ca K series and the Hg L series show the presence of two bows, one evident to unaided eye and one hidden by a black pigment.
Both the areas are characterized by a strong presence of Hg, due to the hypothesized use of cinnabar by Raffaello, and where the two bowls overlap the intensity of Hg increases. It is interesting also to note a very high intensity of K lines of Ca in the hidden bow. This is mainly due to the upper layer of bone black, painted by Raffaello to hide the bow.

3.3 FORS (Fiber Optics Reflectance Spectroscopy)

When a radiation beam impinges onto an object part of the radiation is reflected at the interface air/object, maintaining the same spectral composition of the primary beam, and part penetrates inside the object, where it can be selectively absorbed depending on the chemical and physical nature of the object itself, and then isotropically scattered. When reflectance measurements are performed, both contributions are measured and only the partly absorbed radiation is useful for identification purpose because provides information about the material composition.
Some grouping of the different kinds of electronic transitions is advisable as it can also reflect the electronic structure of the materials under examination. The color of indigo, or in general of the organic pigments, is due to the transitions between delocalized molecular orbitals (DMO), while the charge transfer transitions are responsible for the yellow color of lead chromate or the blue color of Prussian blue. The color of many inorganic pigments as azurite, smalt or verdigris is determined by ligand field transition and the energy band transitions are responsible for the color of a lot of “semiconductor” pigments as cinnabar, litharge, cadmium red. The identification of a pigment is based on having some specific features in the spectrum (the maximum, minimum, inflection points ,shoulders), even if it strongly depends on the presence of mixtures, on the pigment grain size and on painting surface conditions (varnish yellowing, dust darkening and so on).
Progress in fiber optics technology in recent decades has increased the use of in situ instruments to perform reflectance spectroscopy measurements. Diffuse reflectance spectroscopy has mainly three areas of application: the identification of pigments (single or mixture), the measurement of color changes and color-matching. An important contribution to the application of fiber optics to reflectance spectroscopy was given by the researchers at IFAC-CNR in Firenze (Italy). Creating a spectral database, that includes both mixtures and single pigments, they helped in the identification of unknown spectra by means of comparison with those of the reference materials.
An interesting application of FORS is that reported by Poldi and Caglio; in this work they discussed the importance of identifying pigments synthesized and marketed during XX century, by studying phthalocyanine blue and green pigments (i.e. organic polycyclic compounds). This research evidenced that reflectance spectroscopy, performed in the range 200-1000 nm, allows to detect blue and green phthalocyanines, whose spectroscopic features are detected in the wide range 400–850 nm. As a consequence standard portable spectrometers—normally used as colorimeters—operating within the 360–740 nm range can be considered only partially adequate to detect phthalocyanine dyes and pigments.

3.4 Synergy between EDXRF and FORS

Recently non-invasive in situ analyses were performed on seven books of the collection of choir-books with lyrics and scores stored in the old library of the Certosa di Pavia (Italy), considering fourteen illuminated pages in total, painted by four authors. They showed different technical features, but the same materials. Pigments and binders were examined exploiting the synergy between four complementary techniques performed by portable instruments, namely EDXRF (Energy Dispersive X-ray fluorescence), FORS (Fiber Optics Reflectance Spectroscopy), FTIR (Fourier transform infrared spectroscopy) and Micro-Raman spectroscopy.
Intriguing is the case of the use of madder lake for red details, the responsible for the bright fuchsia or magenta color. This pigment was largely used but it is photosensitive, so it usually fades. Instead the choir-books were kept closed most of the time, and so they still hold their original strong shade. It is easily recognizable, for instance by FORS from the band at about 530−540 nm.
The combination of natural ultramarine blue and azurite offers a good example to better understand how EDXRF and FORS can work together. Four blue areas were considered (point 25 on folio 1R from book 822, point 65 on folio 25R from book 821 and point 73 from folio 5V and point 93 on folio 16R from book 814, see fig. 5).
From EDXRF analysis, no chromophore elements were detected for point 65, indicating that a blue pigment based on light element was used. Consistently, FORS spectrum showed the trend of natural ultramarine blue that is mainly composed of Na, Al, Si, O, S. In this case, a monolayer of natural ultramarine can be inferred. The three other points (25, 73 and 93) are characterized by the significant presence of Cu in EDXRF spectra, indicating a probable use of azurite; FORS spectra showed instead different trends, that can be ascribed, respectively, to azurite, natural ultramarine, both pure and combined with azurite. In the first case, the blue color can be supposed to be due to a monolayer of azurite. Results for point 73 reveal instead the presence of a multilayered structure, with ultramarine blue thick layer over azurite. For point 93, the FORS spectrum showed in the infrared region features of both pigments: in fact, the relative reflectance grows up after 700 nm as ultramarine and after 900 nm starts to increase like azurite spectrum. This trend indicates that the spectrum is the combination of ultramarine and azurite: this can be interpreted as a mixture of the two pigments, or, more probably from art historian point of view, a very thin layer (velatura) of lapis over azurite.