The Application of Cathodoluminescence (CL) for the Characterization of Blue Pigments

Abstract

The combined application of Cathodoluminescence (CL) with Scanning Electron Microscopy (SEM) on paintings and painted surfaces has the potential to identify both organic and inorganic pigments on a micrometre or even nanometre scale. Additionally, the combination with Energy-Dispersive Spectrometry (EDS) allows for a more holistic, elemental, and mineralogical characterization of pigments. In addressing the need for the creation of a robust, open access database of characteristic CL spectra of pigments, a large project has been undertaken, focusing primarily on common organic and inorganic pigments. The present paper focuses on the CL characterization of 10 significant blue pigments in pure powder form: cerulean blue, Egyptian blue, Han blue, indigo, lapis lazuli, Maya blue, phthalo blue, vivianite, ultramarine blue, and zirconium blue. The CL spectra present characteristic bands for most of the pigments, allowing their secure identification, especially when combining the results with the EDS analyses. The effect of binding media and of the mixture of different pigments was also studied, via the analysis of mixtures of pigments with oil painted over canvas. Overall, both the binding medium and the mixture of pigments do not appear to create significant differences in the occurring spectra, thus allowing the identification of individual pigments. EDS and RAMAN spectra are included in order to facilitate comparison with other databases.

1. Introduction

Cathodoluminescence (CL) can be described as the optical phenomenon produced when high-energy electrons from a cathode ray tube interact with a solid. The wavelengths of the resulting fluorescence emission are characteristic of the electronic states of the structure of the material, and more specifically of the point defects that affect the crystals (e.g., as atom vacancies, chemical impurities, etc.) [1]. Therefore, the study of CL spectra can allow the determination of the crystallization or genesis of minerals and their mineralization history.

CL has been applied in a vast array of geological materials (see for example [2,3]). In recent years, an increasing number of CL studies have focused on the analysis of archaeological and Cultural Heritage artefacts (i.e., the analysis of quartz for the classification and provenance analysis of pottery [4,5,6,7], the study of natural [8] and man-made glass [9,10,11], the determination of weathering effects on Greek white marble [12], the examination of lime carbonates in plaster from Mexico [13], and provenance studies of Late Neolithic ornaments made of spondylus shells recovered in Hungary [14]).

The application of SEM/CL has proven to be particularly useful for the identification of pigments since the technique can successfully tackle some of the analytical problems which can significantly complicate the analysis of paintings and painted materials [15]. Two significant advantages of the technique are that both organic and inorganic materials can be identified and that the magnification level can reach a micrometre or even a nanometre scale. Additionally, when combined with Energy-Dispersive Spectrometry (EDS), it can lead to a robust, elemental, and/or mineralogical characterization of pigments. The only significant drawback is the need for sampling, although microsamples of extremely small size suffice for the analysis. Overall, the few applications of SEM/CL in the analysis of paintings so far have provided with promising results [15,16,17,18].

The present work is part of a broader project, aiming at the production of a database of characteristic CL spectra of pigments which will facilitate the further application of SEM/CL in the examination of Cultural Heritage materials. The successful application of the technique to the most significant historical white pigments has been previously published by the authors in [15]. More specifically, it was clearly shown that calcite, kaolinite, lead white, zinc oxide, barium sulphate, lithopone and titanium white can be identified via SEM/CL, both in pure form and when mixed with common binding media.

The present paper focuses on 10 common pigments of blue colour. The pigments were selected based on various criteria, aiming to represent the wide array of blue pigments used through the millennia: different chromophores (Co, Cu, Fe, silicates and organic compounds), common or expensive/rare pigments (e.g., lapis lazuli), naturally occurring or synthetic, organic or inorganic, etc. Additionally, all of the selected pigments have been in use for long periods of time and over large geographic areas, thus covering some of the most significant materials available to artists throughout history.

Firstly, the pigments were analysed via SEM/CL in pure powder form. Additionally, EDS and RAMAN analyses were carried out in order to facilitate comparisons with other published databases. Then, the effects of more complex parameters—such as the presence of binding media and/or ground layers, and the mixture of different pigments—were examined via the analysis of especially prepared samples.

2. Brief Overview of Selected Blue Pigments

The blue colour is one of the three fundamental hues in colour theory and occupies a pivotal role within Cultural Heritage. Its profound connections with the portrayal of the sky and the sea, along with its deep-rooted association with religious symbolism, rendered blue pigments indispensable for depicting specific iconic figures such as the Virgin Mary and Christ during the Middle Ages. The demand for blue pigments marked the invention of the first documented synthetic pigment in human history, i.e., Egyptian blue [19]. The present work focused on the analysis of 10 significant blue pigments (namely cerulean blue, Egyptian blue, Han blue, indigo, lapis lazuli, Maya blue, phthalo blue, vivianite, ultramarine blue and zirconium blue); a short overview of each pigment is provided, spanning from their initial utilization to their continued relevance in contemporary times.

2.1. Cerulean Blue

Cerulean blue, also called cobalt cerulean blue or cobalt stannate, was first discovered in 1789 by Höpfner but was apparently largely forgotten until the 1850s–1860s, when it was reintroduced by Messrs G. Rowney and Co. under the name “coeruleum” [19,20,21]. Cerulean blue is a very good drier, with good overall pigment properties. According to the Colour Index, the commercial product usually approximates to 18% CoO, 50% SnO₈, and 32% CaSO₄ [20]. Due to its high cost, it is rarely employed in industrial paints; it is usually produced in small quantities for artistic and ceramic use [22]. It was considered particularly valuable to landscape artists, especially for sky tones, due to its deep blue colour [23]. It has been identified in the palettes of impressionists and in oil paintings of the late 19th century [19].

2.2. Egyptian Blue

Egyptian blue was invented in Egypt during the Fourth Dynasty (around 3100 BCE) and is considered the oldest synthetic pigment. It consists primarily of the mineral cuprorivaite (CaCuSi₄O₁₀), along with some unmelted or partially melted quartz grains. It was extensively utilized in Egypt until the end of the Roman period; its use extended across diverse materials, including stone, plaster, pottery, wood, papyrus, canvas and wall paintings [24]. Following this period, its use gradually declined, with only sporadic appearances in early Middle Ages wall paintings.

The manufacturing process for Egyptian blue emerged as a result of established technical practices, evolving over time. No historical records provide the precise formula employed in ancient times, although experiments initiated in the 19th century have demonstrated that its production involved heating a solution comprising silica, a copper compound (likely malachite), calcium carbonate and sodium carbonate [25]. However, historical specimens of the pigment reveal considerable variability in bulk composition, often containing unreacted synthetic analogue compounds and additional phases derived from the production process [19,24,26]. Due to its variable composition, the analysis of Egyptian blue presents unique challenges and despite the application of advanced analytical techniques, complexities still persist in its analysis and detection [24].

2.3. Han Blue

Han blue, characterized as an isomorphic and chemical analogue of Egyptian blue, distinguishes itself by the incorporation of barium in lieu of calcium within its composition. This unique pigment finds its counterpart in nature in the form of the mineral effenbergite, with both the mineral and Han blue sharing an identical chemical composition, primarily composed of barium copper silicate (BaCuSi₄O₁₀) [27]. The pigment played a very significant role in the decoration of pottery, metallic artefacts and wall paintings of Ancient China. This rich tradition spanned across the Han dynasty (208 BCE−220 AD) and the Warring States Period (475−221 BCE), serving as a testament to the exquisite artistry of the era [19]. Detailed descriptions of the manufacturing procedures can be found elsewhere [19,28].

2.4. Indigo

Indigo is derived from the leaves of various plant species (e.g., Indigofera species native to India, South and Central America, Asia, etc.). The pigment was considered too dark for direct use and was commonly mixed with white pigments [29]. The use of indigo is well documented and spans several centuries. It has been found in Roman wall paintings of the 1st century CE and its use is mentioned in multiple mediaeval manuscripts in Europe. Additionally, indigo has been identified on Mayan pottery and frescoes from the 3rd century, as well as on later examples of Central Asian wall paintings. The pigment is commonly identified in European paintings, from the 15th century in Italy and the Netherlands, until the 19th century in Britain [19].

2.5. Lapis Lazuli

Lapis lazuli is one of the most significant pigments in art history. It is a semi-precious stone, characterized as a fusion of various minerals, typically including calcite, pyrite and most notably lazurite [19]. Lazurite is a cubic sodic-calcic aluminosilicate sulphate mineral [(Na,Ca)8(Al,Si)12O24(S,SO₄)]. It exhibits a remarkably intricate structure, with several known polymorphs, which can coexist even within a single fragment of the mineral. Its intricate chemical makeup remained unknown until the early 19th century [30].

The best-known deposit of lapis lazuli is located in the Kokcha river valley of Afghanistan; the mineral has been mined for over 6 millennia, for use in jewellery and as an ornamental stone [19,30]. The earliest known use of lazurite as a pigment dates back to the 6th to 7th century CE, on the cave temple wall paintings in Bâmiyân, Afghanistan. Since then, the pigment has been identified in China and Persia, while its use exploded in the 14th and 15th centuries in European painting [19]. The cost of the pigment was extremely high, given that it was imported from Afghanistan and had to undergo a difficult preparation process, leading to a selective use by artists for specific parts of their works that they wanted to give prestige to. In the following centuries, its use declined. In 1828, the production of artificial lazurite, or ultramarine blue, was achieved by Jean Baptiste Guimet.

2.6. Maya Blue

Maya blue is a synthetic pigment, invented and used in ancient Mesoamerican cultures spanning from Mexico to Guatemala, and even in colonial sites within the Caribbean region [31]. The pigment is produced by combining the mineral palygorskite [(Mg,Al)2[Si₄O₁₀](OH).4H₂O] with the organic pigment indigo [19]. Eastaugh et al. describe the manufacturing process of Maya blue as involving the heating of a mixture of palygorskite and indigo at approximately 150 °C for two days [19]. However, Sanchez et al. suggest that the production process lacked standardization, with variations in heating time ranging from minutes to hours or days, and temperatures varying between 90–100 °C, 190 °C, and 250–300 °C [31].

Maya blue found extensive application in painting murals, sculptures, ceramics and textiles among the Mayan civilization. Its usage probably also extended to the Toltecs, Mixtecs and the Aztecs [31].

2.7. Phthalo Blue

Phthalo blue is part of a large group of synthetic organic macromolecules (the phthalocyanines), which were widely used as pigments from the third decade of the 20th century onwards. It is believed that by 1981, phthalocyanine pigments accounted for over 90% of all blue and green pigments used by the UK paint industries; however, these pigments have been rarely identified in paintings [19]. Phthalo blue is characterized by some very significant properties (e.g., light fastness, tinting strength, covering power and resistance to alkalis and acids).

2.8. Ultramarine Blue

As mentioned before, ultramarine blue is the synthetic equivalent of the mineral lazurite. Today, the term is used to describe any artificially prepared pigment of similar composition to lazurite. Ultramarine blue is a three-dimensional aluminosilicate lattice with a sodalite structure containing entrapped sodium ions and ionic sulphur groups (Na₄Al₃Si₃S₂O₁₂) [20]. Although the quality of the pigment was initially not very good, its superiority over other blue pigments available during the first part of the 19th century (such as Prussian blue, indigo and azurite) led to its immediate adoption by the painters [32].

2.9. Vivianite

Vivianite is a mineral consisting of hydrous iron phosphate (Fe₃(PO₄)₂.8H₂O) that has been used as a pigment from the Roman period until today [19]. Vivianite is found in the form of delicate white crystals that display a malleable fracture pattern. Upon their exposure to air, these crystals undergo a transformation, adopting a deep blue or blue-green hue [33]. The utilization of the pigment is predominantly associated with regions where the mineral is readily accessible and its use has remained limited [19]. It has been detected in remnants of Roman pigments in Germany, 12th century English wall paintings and in icons spanning from the 11th to the 18th century Moreover, it has been observed in paintings from the 17th and 18th century in Austria [33].

2.10. Zirconium Blue

Zircon is a colourless mineral when it has a pure composition; however, in nature, it commonly appears coloured, due to impurities, crystal defects and other heterogeneities. Zircon can combine with oxygen and silicon, forming as a result several compounds used as pigments of various colours [19]. It should be noted that the Kremer pigment of zirconium blue is called “Zirconium cerulean blue” because of its similarity in colour to cerulean. However, the two pigments are not chemically similar since zirconium blue does not contain cobalt.

3. Materials and Methods

3.1. Samples

Contemporary paints and dry materials often include technological additives to impart specific properties to paints, or inert fillers to reduce the cost; these additives, however, are commonly omitted from the manufacturer’s description [18]. Furthermore, in many instances, the names assigned to pigments primarily reflect their hue rather than their intrinsic composition [19,22]. Taking this into account and with a primary objective of choosing the purest possible pigments for the construction of the database for CL signals, the pigments of Kremer Pigmente GmbH, Germany, were selected. The pigments analysed were the following: Cobalt Cerulean Blue (#45730), Egyptian blue (#10060), HAN-Blue (fine) (#10071), Indigo (genuine) (#36000), Lapis Lazuli (purest) (#10530), Maya blue (light) (#36030), Phthalo Blue (primary) (#23050) and Phthalo Blue (reddish) (#23070), Ultramarine Blue (very dark) (#45000), Vivianite (#104000) and Zirconium cerulean blue (#45400).

Three distinct methods of preparation were employed; it should be highlighted that in all cases, a very small sample was required for analysis (e.g., a few grains of the pigment in powder form), given the high spatial resolution of the SEM: (1) all pigments were analysed in powder form, without undergoing any further preparation, for a more precise detection of the pigment signal; (2) selected pigments were combined with linseed oil and applied to prepared canvases with a chalk ground layer, to detect any variations in pigment behaviour due to interference from other materials; and (3) selected pigments were combined with linseed oil and titanium white (in rutile form), and then applied to prepared canvases with a chalk ground layer to determine the extent to which the obtained signal predominates when the pigment is used in a mixture.

The decision to analyse the pigments in a mixture with a second pigment was made to create the closest possible approximation of a real sample and to determine whether the resulting spectra would beaffected. Titanium white was not chosen for historical accuracy but for comparison purposes and because of its medium cathodoluminescence emission intensity [15]. Linseed oil and pigment samples on canvas were prepared using a ratio of one part oil to four parts pigment. The pigments were then blended with the oil and applied to the canvas with a brush in a thin, even layer. To prepare the pigment and titanium white samples with linseed oil on canvas, a ratio of one part oil, two parts blue pigment, and two parts titanium white were employed. The pigments were combined with the oil and applied to the canvas with a brush, ensuring a thin, even layer. All samples were allowed to air dry at room temperature for a duration of 30 days before proceeding with the analyses.

It is important to note that linseed oil would not be recommended for all the selected blue pigments. However, it was chosen as a common binder to facilitate comparisons. The linseed oil used was oil code #73054, Linseed Oil, cold-pressed, produced by Kremer Pigmente GmbH. The composition of the canvases was evaluated using portable Raman spectroscopy and SEM/EDS analysis.

3.2. Methods

Scanning Electron Microscopy was applied in the Museo Nacional Ciencias Naturales (Madrid, Spain), via a FEI INSPECT ESEM. The SEM resolution at low-vacuum was at 4.0 nm at 30 kV (BSED). Energy-Dispersive Spectrometry (EDS) of the samples was carried out with an energy-dispersive X-ray spectrometer by Oxford Instruments, under an accelerating voltage of 20 KV, using INCA software for the quantification of the data. Precalibration tests of SEM/EDS chemical measurements were performed on internal standards to improve the ZAF correction procedure (Z: atomic number; A: Absorption effect; F: Fluorescence effect).

The SEM setting has MONOCL3 Gatan (CL) detectors, making it possible to work in panchromatic and monochromatic mode with a PA-3 photomultiplier attached to the ESEM. The photomultiplier tube covers a spectral range of 250–850 nm and is more sensitive in the blue parts of the spectrum. A retractable parabolic diamond mirror and a photomultiplier tube were used to collect and amplify the CL signal. No filters were applied to standardize the sensitivity of the photomultiplier tube. The samples were positioned at 16.23 mm beneath the bottom of the CL mirror. The excitation for CL measurements was provided at a 30 kV electron beam. The settings of the analyses were as follows: dwell time, 1.5 s; range, 600 nm; step size, 1.5 nm; PMT VOLTS, 1000–2000; Range (nm), 600 (250 to 850 nm); grating blaze wavelength, 0.15. All CL spectra presented below were the outcomes of single analyses but are considered representative of multiple analyses carried out on different areas of each sample.

Raman analyses were carried out via a BRAVO™ handheld Raman spectrometer (by Bruker Optics GmbH & Co., Ettlingen, Germany). The spectrometer is equipped with two near-infrared excitation lasers (DUO LASER™, wavelengths at 785 nm and 853 nm) and a CCD detector, allowing for a spectral range of 170–3200 cm⁻¹ and a spectral resolution of 10–12 cm⁻¹. The two lasers operate in a patented sequentially shifted mode (SSE™, Sequentially Shifted Excitation), allowing for the mitigation of fluorescence from the samples. The duration of each measurement was typically 180 s. Data acquisition and processing were conducted with the OPUS 8.7.31 software.

4. Results

4.1. Characteristic CL Spectra

Of the 10 blue pigments examined in the present work, only Egyptian blue and Lapis Lazuli have been previously studied via SEM/CL [17,18,34].

Table 1. Representative panchromatic images and CL spectra of the blue pigments, in powder form.

Sample	Panchromatic Image	CL Spectrum
Cerulean blue (Co-Sn oxide)
Vivianite Fe3(PO4)2.8H2O
Zirconium blue (Zr,V)SiO4
Lapis Lazuli (Na,Ca)8 [(SO4,S,Cl)2(AlSiO4)6]
Ultramarine blue (Sodium-aluminium-sulfo-silicate)
Indigo (Indigotin)
Maya blue (Indigo containing magnesium aluminium layered silicate)
Phthalo blue C32H15N8Cu
Egyptian blue CaCuSi4O10
Han blue CuOBaO4SiO2

shows the CL spectra and panchromatic images for each of the studied pigments. The CL panchromatic images can provide an evaluation of the strength of the signal since brighter areas suggest a stronger CL signal; therefore, the homogeneity of the analysed material can also be evaluated. Additionally, pigments which show a strong CL signal will be easier to identify in a real case study, i.e., present in small quantities, mixed with other elements and aged.

The CL analyses were carried out both on individual grains showing the highest CL emission on the panchromatic image and on broader areas of the sample including multiple grains. The CL spectra presented below are representative and reproducible examples of multiple analyses, from diverse areas, unless clearly stated otherwise. A short description of the microstructure of the samples, as well as the chemical composition of the major elements estimated via EDS of the areas corresponding to the CL spectra, is provided below; EDS spectra and BEC images are presented in Table S1 of the Supplementary Materials.

The cerulean blue sample consists of small grains (<20 μm) and contains primarily Co and Sn; significant amounts of Mg and Al are also noted. The sample presents a clear CL spectrum, with a strong band at 312 nm and a broad band at 570 nm. The panchromatic image of the sample shows varied signal strength among different grains. Vivianite presents a broad band at 540 nm; again, the panchromatic image shows highly varied signal strength among grains. Zirconium blue does not have a clear spectrum; however, a small band at 280 nm and a stronger band at 480 nm can probably be used for the identification of the pigment. The latter band can be attributed to dysprosium, a rare earth element that is present in all Zr and creates a well-documented CL band [35].

Regarding lapis lazuli, it should first of all be noted that the EDS analysis suggests the presence of Fe-rich grains (Table S1 of the Supplementary Materials); lapis lazuli is commonly produced as a mixture of the mineral lazurite with calcspar and iron pyrites (FeS₂). The Fe-rich grains do not seem to emit a CL signal, and the CL spectrum presented in Table 1 corresponds to individual lazurite grains (the bright areas on the panchromatic image). The triple band at 384, 425 and 458 nm and the broad band at 766 nm probably relate to the Ca compounds of the mineral. Combined with the strong band at approximately 550 nm, the overall spectrum seems to be characteristic of the pigment. The intensity of the bands is very strong, suggesting that the pigment will be easily identified even when found in a mixture of diverse materials.

Ultramarine contains multiple very small grains (<10 μm), along with a few larger grains (20–30 μm). The examined ultramarine sample is an example of the apparent enhancement of a commercial pigment with compounds that are not historically related. More specifically, EDS analysis suggests the presence of Cu-rich grains, which, however, do not seem to contribute to the CL signal (dark grains on the panchromatic image). Analyses of brighter grains (in which Na, Al, Si, K and S were identified in varying amounts via EDS) produce a low-intensity spectrum with poorly defined bands, which are not always reproducible. Overall, due to the poor quality of the resulting spectra and the unexpected presence of Cu in the pigment, additional analyses are required to determine the CL behaviour of ultramarine.

Indigo contains grains of varying sizes (10–80 μm). Even though it is an organic pigment, its chemical composition suggests the parallel presence of inorganic compounds. The EDS spectra of different grains show large amounts of C (from the organic extract) and varying amounts of Al, Si, Ca and Fe (from the alum used for the production of the pigment). The presence of P can probably be attributed to the organic extract. Several grains were identified as alkali or plagioclase feldspars (Table S1 of the Supplementary Materials). The CL spectrum of the pigment presents multiple bands, some of which could be attributed to the feldspars and other Si compounds of the powder (e.g., the band at 424 nm, perhaps also the band at 646 nm) [36]. However, the rest of the bands are likely due to the organic components of the pigment and could, therefore, be considered characteristic.

Similarly, in the case of Maya blue, some of the bands likely correspond to the Si compounds of the pigment (e.g., bands at 420 and 554 nm). Combined with the sharper bands at 292 and 310 nm, the overall spectrum is probably characteristic of the pigment. It should be highlighted that despite the presence of a common organic compound in indigo and Maya blue, the two pigments can be easily differentiated via their CL spectra.

Phthalo blue contains multiple very small grains (<5 μm), along with a few larger and rounded grains (10–20 μm). The CL analysis of both forms of the Kremer pigment used did not result in a CL spectrum and, therefore, it appears that the pigment cannot be identified with this technique. However, it should be noted that combined with the EDS analysis and the identification of Cu, the lack of a CL signal could be used as an indication of the presence of phthalo blue.

Egyptian blue contains relatively large (20–80 μm) angular grains and the EDS analysis suggests the presence of Cu, Si and Ca. Its panchromatic image shows the presence of multiple large grains emitting strong CL signals. The spectrum shows two smaller bands at 382 and 552 nm and a strong band at approximately 860 nm. Kadikova et al. [18] analysed a similar Egyptian blue sample, resulting in a spectrum with a singular strong band at approximately 873 nm. The reasons for this differentiation need to be more thoroughly understood, but in any case, it appears that the pigment can be easily identified by the strong band in the 860-870 nm region.

Finally, Han blue contains grains of an approximately 30 μm diameter and consists primarily of Ba, Si and Cu, as expected. The sample presents a broad band at 430 nm and a characteristic sharp band at 870 nm. As also evident by the panchromatic image, the overall emission of the CL signal is relatively low in the sample.

4.2. Effect of Binding Media, Pigment Mixture and Multiple Layers

Previous experiments with mixtures of pigments with different binding media (egg yolk, linseed, walnut and poppy seed oil) have suggested that none of these common binding media produces a significant CL signal and, thus, they cannot hinder the identification of a pigment via CL [15]. Furthermore, it was shown that the CL signal from specific pigments can be distinguished even in cases of multiple painted layers and mixed pigments [11]. To further explore the effect of these parameters, various mixtures of pigments with linseed oil were painted over a canvas (

Table 2. Description of different mixtures of pigments painted over canvas.

Sample	Blue Pigment	White Pigment	Binding Medium	Canvas (with Chalk)
M1	-	-	-	X
M2	-	Titanium white	Linseed oil	X
M3	Lapis lazuli	-	Linseed oil	X
M4	Lapis lazuli	Titanium white	Linseed oil	X
M5	Indigo	-	Linseed oil	X
M6	Indigo	Titanium white	Linseed oil	X
M7	Maya blue	-	Linseed oil	X
M8	Maya blue	Titanium white	Linseed oil	X
M9	Egyptian blue	-	Linseed oil	X
M10	Egyptian blue	Titanium white	Linseed oil	X
M11	Han blue	-	Linseed oil	X
M12	Han blue	Titanium white	Linseed oil	X

) and then analysed with SEM/CL. A sample of clear canvas and a sample with only titanium white (in rutile form) were analysed for comparative reasons (Table S2 of Supplementary Materials).

In all cases, the samples were also analysed with EDS and Raman (Tables S3 and S4 of the Supplementary Materials). It is interesting to note that in most cases, the Raman spectra are very complex, as apart from the pigments, chalk (used as a ground layer on the canvas) also produces several bands (Tables S4 and S5 of the Supplementary Materials); the interpretation of the spectra is, therefore, significantly hindered.

Based on the representative spectra presented in Table S2 of the Supplementary Materials, the unpainted sample of the canvas (M1) presents a characteristic CL signal, with a broad band at 422 nm (probably due to Ca compounds) and a strong band at 614 nm (probably due to Mn²⁺). Despite the high intensity especially of the latter band and the relatively thin layer of the applied paint, the canvas signal is only noted on one of the analysed samples (M2). This clearly suggests that the technique is not easily affected by the presence of multiple layers of paint or ground layers unless the layers are particularly thin.

The titanium white sample (M2), in the form of rutile, presents its characteristic bands at 436 and 475 nm [15], although in this case, the canvas band at 614 nm is also seen. However, the pigment cannot be identified in most of the mixtures with the blue pigments. This is not surprising, as it is well documented that titanium white produces a low CL signal [15,18].

Samples M5 and M6 (mixtures of indigo) are the only samples that do not produce a clear CL spectrum, even though the pigment presented characteristic bands with high intensity when analysed in pure powder form. However, the spectra of the two mixtures are identical to the spectra produced by the plain canvas (M1) and titanium white (M2), respectively. This could be attributed to a smaller thickness of the painted layers, although more analyses are required to better understand the absence of CL single from indigo.

All other mixtures present identical spectra with the respective pigments in powder form; linseed oil does not seem to produce any CL signal and the presence of either titanium white or chalk (from the canvas) is not noted. In most cases, the bands of the two mixtures present heightened intensity compared to the respective pigment in powder form. This could be explained by the acquisition process of the spectra: in the pure pigments, analyses were carried out under large magnification, focusing on specific grains or small homogeneous areas, whereas in the painted samples, the analyses were taken from larger areas, thus accumulating significantly more signal.

For example, the three forms of lapis lazuli (in powder form, mixed with linseed oil, and mixed with titanium white and linseed oil) are presented in Figure 1. As is clearly shown, all samples show the same characteristic bands, despite the presence of the other compounds. This pattern is very significant, as it clearly highlights the potential of the technique in identifying pigments, despite the complex nature of the materials present in paintings and other painted materials.

5. Discussion

The combined application of CL with the traditional technique of SEM/EDS has been shown to have significant potential as a powerful new tool for the study of paintings and other painted materials. EDS allows detailed elemental analysis, whereas CL offers the additional possibility of investigating aspects of the electronic states and defects of the structure, thereby providing a more specific mineralogical determination of the analysed samples. Moreover, elemental and luminescence mapping can be carried out in parallel, focusing on the same areas, under the high spatial resolution achieved through SEM, thus leading to a quick and thorough examination of a sample. Furthermore, the CL signal is emitted by both inorganic and organic materials, which is very significant for the analysis of the highly varied materials used as pigments.

Building a CL database, at least for the most common and significant pigments, is a necessary first step in order to apply the technique for the study of paintings more extensively. The results described in the present paper, focusing on a selection of 10 οf the most significant historical blue pigments (cerulean blue, Egyptian blue, Han blue, indigo, lapis lazuli, Maya blue, phthalo blue, vivianite, ultramarine and zirconium blue), indicate that CL can be successfully applied for their identification in most cases, especially via the combined application of the EDS analysis. In particular the pigments of cerulean blue, vivianite, lapis lazuli, Egyptian blue, indigo and Maya blue seem to be easily identified; the significant differences noted in the CL spectra of the two latter pigments are particularly interesting since indigo and Maya blue contain a common organic compound, thereby making their differentiation more difficult with other techniques.

It is important to highlight that pigments could be successfully identified not only in their pure powder form but also when mixed with binding media and/or other pigments and painted over a canvas, further supporting the robustness of the technique. The analysis of additional pigments and dyes is currently underway, aiming to form the basis for a detailed, open access database which will include the elemental and spectroscopic information of the most significant historical pigments.