Spectroscopic characterization of sixteenth century panel painting references using Raman, surface-enhanced Raman spectroscopy and helium-Raman system for in situ analysis of Ibero-American Colonial paintings

Abstract

Colonial panel paintings constitute an essential part of Latin-American cultural heritage. Their study is vital for understanding the manufacturing process, including its evolution in history, as well as its authorship, dating and other information significant to art history and conservation purposes. Raman spectroscopy supplies a non-destructive characterization tool, which can be implemented for in situ analysis, via portable equipment. Specific methodologies must be developed, comprising the elaboration of reference panel paintings using techniques and materials similar to those of the analysed period, as well as the determination of the best analysis conditions for different pigments and ground preparations. In order to do so, Raman spectroscopy at 532, 785 and 1064 nm, surface-enhanced Raman spectroscopy (SERS) and a helium-Raman system were applied to a panel painting reference, in combination with X-ray fluorescence analysis. We were able to establish the analysis conditions for a number of sixteenth century pigments and dyes, and other relevant components of panel paintings from this period, 1064 nm Raman and SERS being the most successful. The acquired spectra contain valuable specific information for their identification and they conform a very useful database that can be applied to the analysis of Ibero-American Colonial paintings.

This article is part of the themed issue ‘Raman spectroscopy in art and archaeology’.

1. Introduction

Wood panel and easel paintings are very important components of Mexican and Latin-American cultural heritage. During the second half of the sixteenth century, the European artists who arrived in New Spain were commissioned to participate in the conformation of complex programmes of images oriented to the decoration of the main liturgical spaces. Their artistic activity supported the mission of the ecclesiastical regime and, at the same time, influenced the local context with novel artistic proposals based on formulae and ideas that had their roots in the art of the Renaissance. Paintings, drawings and engravings accompanied the artists and patrons' committees as agents, in a process where images had an important role within the cultural conformation of New Spain as a newly formed entity.

Flemish, Spanish and Italian artistic traditions converged through the proposals and pictorial solutions settled at New Spain's workshops established by the European artists [1]. In their programmes, it is not rare to find a mixture of techniques or artistic innovations, developed as a form to adapt the commission to the local requirements. This combination of influences gave a particular character to the new images, not only at the iconographic level but also at their material dimension. In this sense, it is considered important to understand the materials and painting techniques of sixteenth-century panel paintings and altarpieces through the means of experimental reproductions.

There are some characteristic features that help distinguish a painting belonging to a certain pictorial tradition from the point of view of its material composition, specifically by means of the ground preparations, layer structure and colour preferences. An interesting approach in the study of the artistic context of New Spain painting is the recognition of the traditions involved within its technological process but there are other concerns, more profound, on the possibilities of artistic innovations related to the use of local materials and the mixture of traditions in the same work of art.

A painting is a very complex object and in order to acquire relevant information about its composition and manufacturing techniques, the use of specific and non-destructive analytical techniques is often vital. The results yielded by such techniques are very valuable for understanding the adaptation of European painting schools to the New World and help to establish the historical changes, modifications and even alterations of the object. As such, these analyses can lead to crucial information for restoration and conservation projects [2]. X-ray fluorescence (XRF) and Raman spectroscopies are suitable techniques for the analysis of paintings, as they fulfil most of the requirements for these kinds of studies.

To acquire relevant information regarding the techniques and materials found in sixteenth century pictorial art in New Spain, a research project focused on the creation of experimental reproductions of panel paintings was developed [3] by an interdisciplinary group from the Laboratory of Artwork Diagnosis of the Aesthetics Research Institute of UNAM. For this project, a historical research campaign based on art treatises and literature specialized in the technical examination of Flemish, Italian and Spanish paintings was conducted, in order to support the creation of a set of painting references where different materials were tested on grounds and paint layers. The pigments selected for this study have a chemical composition corresponding to the materials used in this period.

The mixture preparations and the application of paint layers were guided by historical mentions in European art treatises about the handling properties of pigments and binders, and relied on the experience of a group of artists in order to obtain trustworthy and consistent references, as close as possible to an original artwork from this period [3]. These reference painting materials have been characterized using a wide variety of spectroscopic techniques to optimize the experimental conditions and to test the feasibility of the application of these analytical techniques.

Raman spectroscopy is a molecular vibrational technique that is based in the dispersion of a laser beam by the molecules of the analysed material. Because of its high specificity and the possibility to perform direct analysis of objects, this spectroscopic technique has become a powerful tool in the study of cultural heritage and has been successfully applied to the analysis of pigments [4], documents [5–8] and mural paintings [9], among others. Nowadays, the availability of portable Raman spectrometers permits the in situ study of objects in museums, archaeological sites, etc. Raman analysis together with other spectroscopic techniques like XRF can yield a complete characterization of the material composition of an object.

Two phenomena associated with Raman spectroscopy hinder its effectiveness: the low intensity of the signal—related to the probability of occurrence of the Raman effect—and the not uncommon presence of a luminescence background, due to the excitation of electronic states of organic and inorganic materials [10] used in the elaboration of artworks, such as dyes, resins, binder or varnish. For this reason, Raman spectroscopy is a difficult technique to apply, particularly for in situ analysis. Surface-enhanced Raman spectroscopy (SERS) is a related technique that uses metal nanostructures to enhance the Raman signal and quench the luminescence background of the materials. The technique is based on the interaction of light with metal nanostructures and the preparation of SERS metallic substrates is a constant growing field of research. The most common methods include the elaboration of metallic colloids (silver, gold or copper), where the chemical reduction of the metal ions is the most used [11]. SERS has been successfully applied to the study of dyes [12–14] and resins [15].

A less addressed subject in Raman spectroscopy is the heating effect from the laser on the analysed surfaces, which becomes important when the laser power is above certain limits [16–19]. In a previous work [20], we reported the use of a helium flux (HERAS) to avoid these thermal effects, inspired by the application of He to reduce the ion and X-ray energy loss and to decrease the damage caused by the ion irradiation heating on the sample [21,22]. A later development of the same technique allows the simultaneous use of a microscope.

It is important to establish the conditions for an effective and safe application of Raman spectroscopy and its related techniques to the analysis of sixteenth century panel paintings. In order to do so, one of the above-mentioned experimental reproductions was analysed using various Raman techniques. Raman measurements were carried out by means of three different Raman laser wavelengths, as well as SERS and HERAS. The goal was to determine the best parameters for the study of original paintings and generate a Raman database of Novo-Hispanic art materials. XRF analysis was performed to complement and confirm the Raman findings.

2. Experimental procedure

Analyses were performed on a painting trial that was prepared following historical sources and art treatises. A ground layer of gypsum was applied over a panel made of pinewood. The surface was sealed with a thin coat of linseed oil and received a green coloured priming (made of lead white, green earth, azurite, ochre and black). Twenty-five pigments (table 1) were applied, without further characterization, using linseed oil as a binding medium, as part of a project from the Laboratory of Artwork Diagnosis of the Aesthetics Research Institute of UNAM [3,23].

Table 1.

Pigments and dyes used to prepare the panel painting reference. (Online version in colour.)

Pure pigments were also analysed for comparison reasons. The list of pure pigments includes yellow ochre, orpiment, lead–tin yellow type II, weld dye (Sennelier), weld dye (Zecchi), fawn ochre, burnt umber Cyprian, dark ochre, dark burnt Sienna, asphalt, vermilion, red iron oxide, minium, cochineal, alizarin, madder lake, azurite, smalt, ultramarine blue, indigo, copper resinate, malachite, green earth, carbon black and noir de vigne. The spectra were compared with those reported in the literature and databases such as RRUFF [24].

Raman spectra were acquired using three spectrometers: i-Raman Plus portable spectrometer (B&W Tek) with a 532 nm laser beam, 175–4000 cm⁻¹ spectral range, CCD array detector and variable laser power; an Inspector Raman portable spectrometer (Delta Nu) with a 785 nm laser beam, 200–2000 cm⁻¹ spectral range, CCD array detector and a 20 mW fixed laser power; and an i-Raman EX portable spectrometer (B&W Tek), with a 1064 nm laser beam, 175–2500 cm⁻¹ spectral range, InGaAs array detector and variable laser power. Raman measurements were carried out in situ without sample preparation. The auto create baseline correction (with 10 points) from OriginLab was used on all of the spectra, and some of the spectra acquired with the 1064 nm laser were also smoothed with the Savitzky–Golay method (7 points of window and a second order polynomial).

SERS analysis was performed with the 785 nm Inspector Raman. Citrate-reduced silver colloids were prepared using a variation of the Lee–Meisel method [25]. The size of these nanoparticles was determined via transmission electron microscope images processed with Image-J software, obtaining particle sizes between 5 and 100 nm with a distribution centred on 24 nm. The absorption maximum of the colloid was around 415 nm in the UV–visible characterization. Silver colloid (20 μl) was mixed with 2, 5 or 10 μl of KNO₃ to induce the agglomeration of nanoparticles, and 3 μl of the mixture were deposited on the surface of the paint with a micropipette.

Helium Raman system (HERAS) measurements were achieved using the 785 nm Inspector Raman. A helium device was fixed on the spectrometer as previously reported [20], but for this research the helium device was adapted to the microscope, so that the Helium flux impinges directly on the area where the laser is focused. The helium flux was regulated by means of a gas mass flux regulator (Mass-Trak, Sierra Instruments) with a constant flux of 5 SLPM (standard litres per minute; figure 1).

Figure 1.

Helium flux device for HERAS measurements. A copper tube directs the helium flux to the area of analysis during the measurements.

XRF analyses were performed with SANDRA, a portable equipment developed by our group [26]. XRF analysis was carried out using an Amp-Tek X-123 SDD detector and a Mo X-ray tube operating at 70 mA and 45 kV, with a 90 s acquisition time.

3. Results

The panel painting reproduction was fully analysed using the three Raman spectrometers, SERS and XRF. HERAS was applied only when the analysis region presented some damage due to the laser. Raman spectra of the panel gave information about the pigments, the ground layer, the priming and the binder. Common features were: a band around 1006 cm⁻¹, corresponding to the strongest vibration of calcium sulfate, a band at 1055 cm⁻¹, associated with the lead white [27] priming and bands around 900 cm⁻¹, 1250 cm⁻¹, 1300 cm⁻¹, 1440 cm⁻¹, which correspond to the previously reported Raman spectrum of linseed oil binder [28].

In general, the quality of the Raman spectra obtained with each technique—intended as a high signal to noise ratio and the presence of specific vibrational bands—varied as a function of the pigment analysed. Although a clear tendency was not observed, three cases can be distinguished. For some of the analysed references a clear improvement of the signal was noticed when the laser wavelength was changed from 532 to 1064 nm (figure 2). As an example, the 532 nm laser produced a very intense luminescence background on the weld dye reference, and no bands from the dye, the priming or the ground can be observed in the spectrum. Less luminescence was present when the 785 nm laser was applied, but only the signal from the white lead priming can be clearly detected. The near infrared laser at 1064 nm gave the best quality spectrum in this case, as the bands from gypsum (1006 cm⁻¹), white lead (1055 cm⁻¹), luteolin (1360 and 1594 cm⁻¹) and probably the binding oil (1300 and 1450 cm⁻¹) are clearly observed. Although this behaviour was expected for most of the reference panel, the low intensity associated with the 1064 nm laser led to very noisy spectra in some cases. This problem was partially solved with the SERS approach, which allowed the correct identification of the materials present in a fraction of the painted references where no useful spectra could be acquired with the 532, 785 and 1064 nm lasers.

Figure 2.

Raman spectra of the weld dye reference (Reseda luteola) acquired with different laser wavelengths: 532 nm (a), 785 nm (b) and 1064 nm (c).

A particular case of SERS effectiveness was the indigo reference (Y4-20, figure 3). The 532 nm laser yielded a highly luminescent spectrum, while some spectral features from indigo are visible in both the 785 and 1064 nm spectra (figure 3b,d). However, the 785 nm spectrum of indigo is still affected by strong luminescence while the 1064 nm spectrum shows very weak bands, which made the identification difficult. By contrast, the SERS spectrum acquired with the 785 nm laser shows sharp and intense bands of indigo, even if not all the luminescence was quenched. Similar cases were observed in about half of the analysed references.

Figure 3.

Raman spectra of the painted reference of indigo (Y4-20) acquired using 532 nm (a), 785 nm (b), 785 nm/SERS (c) and 1064 nm (d).

From the 25 painted references, 15 underwent laser-related damages, the areas painted with earth and copper pigments being the most sensitive ones. With a helium flux of 5 SLPM, it was possible to reduce or avoid damages on the surface, and in some cases the background in the spectra was also reduced. Figure 4 shows the Raman spectra of two areas painted with yellow ochre (Y4-01, figure 4a,b) and hematite (Y4-12, figure 4c,d). The vibrational bands of goethite (299, 389 and 548 cm⁻¹) are visible in the HERAS analysis (figure 4b) of yellow ochre, in contrast with the normal Raman spectrum where the luminescence impedes the detection of any such bands. In the second case, Y4-12, even when the Raman spectrum shows two characteristic features of hematite at around 294 and 408 cm⁻¹, the low signal to noise ratio makes the identification challenging. The application of helium led to an increase of the signal to noise ratio and four hematite bands are clearly observed at 230, 294, 408 and 606 cm⁻¹.

Figure 4.

Comparison between conventional Raman and helium Raman system (HERAS) of yellow ochre (Y4-01: a—Raman; b—HERAS) and hematite (Y4-12: c—Raman; d—HERAS).

Besides the fact that a helium jet protects the object from heating damage, the reduction of the luminescence background on the Raman spectra is a very desirable side effect. No clear explanation has yet been formulated to explain this reduction in luminescence background. Two processes have been proposed. The first one is that helium creates an inert atmosphere, avoiding sample combustion together with its associated light emission. The second one is that helium cools the analysed surface, reducing the production of fluorescent species from combustion, reductive and oxidative processes, laser-induced degradation or photochemical reactions [20]. Asphalt (Y4-10), cochineal (Y4-14), indigo (Y4-20) and carbon black (Y4-24) were other pigments that suffered laser-induced damages.

An additional remark is needed regarding the raw materials used to prepare the reference panel. Although all the pigments and dyes were acquired from well-known manufacturers and distributors of artistic materials, we cannot be completely certain of their composition, as they often lack a full data sheet. In order to partially overcome this, XRF was used as a complementary technique. XRF was useful for confirming the identification of inorganic pigments, along with elements present in the priming and ground layer, e.g. lead in the priming and calcium from gypsum in the ground layer. In addition, the pure pigments were analysed with Raman spectroscopy and compared with references from the literature, although the resulting spectra are not shown (see electronic supplementary material).

The following sections detail the results achieved for each painted square of the reproduction organized by colour, showing the best Raman, HERAS or SERS spectra when applicable and always complemented by the XRF information. However, for some of the references no useful spectra could be acquired, in spite of the application of all the Raman alternatives discussed before.

(a). Yellow colours

The first row in the painting reproduction corresponds to yellow pigments. The pigments used were yellow ochre (Y4-01), orpiment (Y4-02), lead–tin yellow type II (Y4-03) and two weld dyes (Reseda luteola) from Sennelier (Y4-04) and Zecchi (Y4-05).

The best Raman spectra of the yellow pigments were acquired using SERS and the 1064 nm laser beam Raman spectrometer. Figure 5 shows the Raman spectra of yellow pigments, in particular the SERS spectrum of Y4-01 shows bands at 299, 389, 548 cm⁻¹ that correspond to yellow ochre (goethite). The Y4-02 Raman spectrum only shows strong bands at 205, 308 and 357 cm⁻¹, which correspond to orpiment. The band at 327 cm⁻¹ in the SERS spectrum of Y4-03 corresponds to lead–tin yellow type II [29]. The XRF measurements were in agreement with Raman analysis in the sense that iron was very intense in the yellow ochre (Y4-01), arsenic in the orpiment area (Y4-02) and tin in the lead–tin yellow type II of square Y4-03, together with the overall presence of lead from the priming and calcium from the ground layer.

Figure 5.

Yellow references analysis: (a) 785 nm SERS of yellow ochre; (b) 1064 nm Raman of orpiment; (c) 785 nm SERS of lead–tin yellow type II; (d) 1064 nm Raman of weld dye (Sennelier) and (e) 1064 nm Raman of weld dye (Zecchi).

The two areas painted with weld dye were difficult to analyse because of the strong luminescence of the organic dye and only with the 1064 nm Raman spectrometer was it possible to obtain information. The spectra show bands at 1360 and 1594 cm⁻¹ which correspond to the v(C=O) and v(C–O) vibrations of the luteolin molecule, respectively [30]. In addition, the XRF measurements of weld dye areas were useful to distinguish between them, as a high amount of potassium was found in the weld from Zecchi.

(b). Ochre colours

Pigments in the row of the ochre colours are fawn ochre (Y4-06), burnt umber Cyprian (Y4-07), dark ochre (Y4-08), dark burnt Sienna (Y4-09) and asphalt (Y4-10). Ochre colours were difficult to characterize because of the strong luminescence they exhibit with the 532 and 785 nm Raman spectrometers and owing to the low energy of the 1064 nm laser. Only with SERS was it possible to obtain spectra of two of the five ochre areas. Figure 6 shows the SERS spectra of fawn ochre (Y4-06) and dark burnt Sienna (Y4-09), where the band at 391 cm⁻¹ corresponds to the strongest band of goethite while the 294, 410 and 615 cm⁻¹ bands have been reported for hematite [31]. In addition, the SERS spectra of Y4-06 and Y4-09 exhibit bands in the 1051–1600 cm⁻¹ region, which correspond to the priming and the binder that have been described before. The spectra of the area painted with burnt umber Cyprian (Y4-07), dark ochre (Y4-08) and asphalt (Y4-10) did not provide any information with the application of all of the Raman alternatives presented. The XRF measurements showed that all the ochre pigments (except asphalt) contain iron.

Figure 6.

Ochre pigments analysis: 785 nm SERS spectra of (a) fawn ochre (Y4-06) and (b) dark burnt Sienna (Y4-09).

(c). Red colours

Pigments used in the red areas of the panel were vermilion (Y4-11), red iron oxide (Y4-12), minium (Y4-13), cochineal (Y4-14), alizarin (Y4-15) and madder lake (Y4-16). Some of the red pigments in the reproduction were easy to analyse and it was not necessary to use SERS, e.g. vermilion (figure 7) and minium. On the other hand, SERS was necessary for pigments like red iron oxide and alizarin. The areas painted with cochineal (Y4-14) and madder lake (Y4-16) present the bands reported by Burgio et al. [27] for carminic acid, where the main difference between the dyes is the high intensity of the 467 and 1692 cm⁻¹ bands in madder lake. Although these results were expected for cochineal, madder consists of a varying mix of alizarin and purpurin, depending on the dye source, and should yield a different Raman spectrum.

Figure 7.

Red pigments analysis: (a) 785 nm Raman of vermilion (Y4-11); (b) 785 nm SERS of Y4-12 (red iron oxide); (c) 785 nm Raman of minium (Y4-13); (d) 1064 nm Raman of cochineal (Y4-14); (e) 785 nm SERS of alizarin (Y4-15); and (f) 1064 nm Raman of madder lake (Y4-16).

A similar case occurred for the Y4-15 reference, as the SERS spectrum does not correspond to those reported for alizarin [32]. The XRF analysis showed a high amount of Ba, which suggests the presence of a different dye. Barium salts of naphthoic acid derivatives—specifically the barium salt of 4-(o-sulfo-p-tolylazo)-3-hydroxy-2-naphthoic acid, known as D & C red 6 barium lake or pigment red 57—can be found as an ingredient of cosmetics and its reported spectrum [33] matches the one we acquired for the alizarin reference. In addition, XRF complemented the Raman identification of the inorganic red pigments vermilion (Y4-11) and red iron oxide (Y4-12) based on the presence of mercury and iron, respectively.

(d). Blue colours

The pigments used in the blue row are azurite (Y4-17), smalt (Y4-18), ultramarine blue (Y4-19) and indigo (Y4-20). The Raman spectrum of Y4-17 could only be acquired with the 1064 nm spectrometer (figure 8) where the band at 405 cm⁻¹ is the most intense in azurite; a copper mineral was confirmed by the XRF measurements. G and D bands of carbon were also observed at 1302 and 1597 cm⁻¹.

Figure 8.

Analysis of the blue areas: (a) 1064 nm Raman of azurite (Y4-17); 785 nm SERS of (b) ultramarine blue (Y4-19) and (c) indigo (Y4-20).

SERS was useful in the identification of cobalt blue and indigo in the Y4-19 and Y4-20 region respectively. It was not possible to characterize the Y4-18 reference by Raman spectroscopy, but the XRF analysis of this area shows the presence of Co, K and Si, characteristic elements of smalt.

(e). Green and black colours

Copper resinate (Y4-21), malachite (Y4-22) and green earth (Y4-23) were employed in the green areas, while the black pigments were carbon black (Y4-24) and noir de vigne (Y4-25). The use of 532, 785 and 1064 nm lasers proved unsuccessful for the identification of green pigments.

The application of SERS yielded good results only in one of the three green areas, assigned to the green earth. However, the Y4-23 spectrum contains bands at 1190, 1289 and 1545 cm⁻¹ (figure 9), which may be related to the presence of phthalocyanine green [27]. In fact, the XRF spectra of this area revealed the occurrence of Cu, also related to this pigment. Both findings indicate that some phthalocyanine green was probably mixed with the green earth.

Figure 9.

Green and black areas measurements: SERS spectra of (a) green earth (Y4-23) and (b) carbon black (Y4-24).

The identification of copper resinate (Y4-21) and malachite (Y4-22) was achieved only by XRF. Carbon black from Y4-24 area was easy to identify by SERS, as observed in the spectrum of figure 9, where the bands at 1327 cm⁻¹ and 1595 cm⁻¹ can be assigned to the G and D bands of carbon, respectively. The noir de vigne pigment (Y4-25) was identified using XRF owing to the high amount of iron determined in agreement with previous examination [34].

(f). Final remarks

Table 2 summarizes the results of the application of Raman spectroscopy, using three different laser wavelengths, SERS and HERAS to the panel painting reference. For 72% of the analysed squares, at least one of the five alternatives led to good results, while only in two cases all three Raman wavelengths and SERS produced useful spectra. We were unable to obtain relevant information for seven of the analysed materials (burnt umber Cyprian, dark ochre, asphalt, smalt, copper resinate, malachite and noir de vigne).

Table 2.

Results summary.

reference material	532 nm Raman	785 nm Raman	1064 nm Raman	785 nm SERS	785 nm HERAS	laser damageb	major elements detected by XRF
yellow ochre			388 vw	299 m,	299 sh,	•	Pb, Fe, Ca, Ba
				389 s,	389 s,
				548 w	548 w
orpiment	294 w,	300 sh,	205 w,	300 sh,			Pb, As, Fe, S,
	311 w,	312 s,	308 s,	312 s,
	353 s	360 vs	357 vs	360 vs
lead–tin yellow type II		327 sh	151 vw	327 sh			Pb, Fe, Sn, Ca
weld dye, Sennelier			1360 m,				Pb, Fe, Ca
			1594 m
weld dye, Zecchi			1360 m,				Pb, K, Fe, Ca
			1594 m
fawn ochre				391 w,	1193 sh,	•	Pb, Fe, Ca
				1193 m,	1284 sh
				1284 m,
				1556 w
burnt umber Cyprian						•	Pb, Fe, Ca, Cu
dark ochre						•	Pb, Fe, Ca, Mn
dark burnt Sienna				294 sh,		•	Pb, Fe, Ca, K
				410 w,
				615 w,
				1196 s,
				1282 s,
				1551 m
asphalt						•	Pb, Fe, Ca
vermilion		253 vs,	256 vs,	253 vs,			Hg, Pb, As, S
		355 m	345 m	355 m
red iron oxide			229 m,	230 vs,	294 s,	•	Fe, Pb, Ca
			296 s,	294 vs,	408 w,
			408 s,	408 m,	606 w
			612 m	606 m
minium	475 w,	317 sh,	120 vs,	317 sh,			Pb, Cl, Fe
	546 s	397 s,	229 vw,	397 s,
		549 vs	309 vw,	549 vs
			388 vw,
			549 vw
cochineal			1313 s,			•	Pb, Fe, Ca
			1448 m
alizarina		500 sh,	1129 sh,	500 sh,			Pb, Fe, Ca, Ba
		747 sh,	1364 s,	747 sh,
		964 m,	1491 m,	964 m,
		1127 sh,	1605 m	1127 sh,
		1181 s,		1181 s,
		1231 s,		1231 s,
		1268 w,		1268 w,
		1368 vs,		1368 vs,
		1490 vs,		1490 vs,
		1600 vs		1600 vs
madder lakea			467 w,				Pb, Fe, Ca
			1315 s,
			1445 w,
			1692 m
azurite			405 m			•	Cu, Pb, Fe, Ca
smalt							Pb, Co, K, Fe
ultramarine blue	544 w	544 vw		544 w			Pb, Ca, Fe, K
indigo			542 m,	546 m,	546 vw,	•	Pb, Ca, Fe
			1538 s	600 sh,	1310 w,
				753 w,	1460 sh,
				1227 sh,	1572 s
				1310 w,
				1362 vw,
				1460 w,
				1573 vs
copper resinate						•	Cu, Pb, Fe
malachite						•	Cu, Pb, Fe
green earth				1190 m,	1190 w,	•	Pb, Fe, Ca, Ti, Cu
				1289 m,	1289 w,
				1545 m	1545 w
carbon black			1293 vw,	1327 vs,	1327 s,	•	Pb, Ca, Fe
			1598 vw	1595 vs	1595 s
noir de vigne						•	Pb, Fe, Ca

^aDiffers from the expected composition.

^bWavelength associated with laser damage is 785 nm because of the power in this spectrometer is fixed at 20 mW.

Intensity vw, very weak; w, weak; m, medium; s, strong; vs, very strong; sh, shoulder.

In spite of the low signal intensity, 1064 nm Raman proved the most effective technique, allowing the acquisition of good quality spectra in fourteen cases. It was particularly useful in the cases of weld dye, madder lake, cochineal and azurite, where the rest of the techniques gave poor results. Nevertheless, it failed to provide good data for the ochre and green colours.

SERS was almost as effective as the 1064 nm Raman, as it yielded useful information for about half of the analysed areas, including all the colour groups. It provided the best results in three cases of iron pigments: fawn ochre, dark burnt Sienna and green earth. The main handicap of SERS is that the metal colloid must be in close contact with the analyte in order to produce the signal enhancement, which limits its application in the case of an original artwork. Also, some compounds used in the elaboration of the colloidal substrate could provide misleading information, such as potassium nitrate (KNO₃) which is used in the nanoparticle agglomeration. Raman spectra from KNO₃ present a strong band near 1055 cm⁻¹ which can be misinterpreted as lead white. Another band is produced by the nanoparticle colloidal silver at 240 cm⁻¹ that can sometimes be detected in SERS analysis.

Resolution and other instrument characteristics of portable spectrometers are usually lower than those of bench-top equipment and this may result in slightly different spectra. Taking this fact into account, the spectra of the panel paintings were always compared with the spectra of the pure pigments and dyes, and to those reported in the literature, in order to ensure correct interpretation.

As discussed before, almost two-thirds of the painted areas were damaged to some extent by the 20 mW 785 nm laser. Of those, the application of HERAS avoided or reduced the damages in all cases, but provided good quality information only in six cases: yellow ochre, fawn ochre, red iron oxide, indigo, green earth and carbon black.

4. Conclusion

In situ studies of a painting are difficult because of its complex composition and the numerous materials used in its elaboration. Raman spectroscopy provided information on the pigments as well as the binder, the ground layer and the priming used in the reference colonial panel painting analysed. When possible, a complete in situ Raman study of a painting must be carried out using different wavelength lasers.

In this research, there was no clear tendency as to which is the best wavelength for each pigment, though 1064 nm Raman seemed adequate for a higher number of materials. SERS approach provided good results but is still an invasive technique, difficult to apply to original artwork. HERAS technique avoided the laser damage on the reference panel painting and, in some cases, e.g. yellow ochre and hematite, the helium flux improved the spectra quality. XRF measurements complemented the identification of the inorganic pigments and provided useful information for some of the organic materials analysed. The acquired spectroscopic data of sixteenth century panel painting reproduction presented here are now available as a reference for the study of Ibero-American Colonial paintings.

Data accessibility

Supporting data can be accessed via Dropbox, at the following link: https://www.dropbox.com/sh/g0r3kyr0asp595i/AACspBKQoD9YzEOC6f1ShzBna?dl=0.

Authors' contributions

M.A.G.-B.: experiment conception and design, acquisition, analysis and interpretation of Raman, SERS and HERAS spectra, analysis and interpretation of XRF spectra, article drafting and revision; E.C.-G.: experiment conception and design, preparation of SERS substrate, analysis and interpretation of Raman, SERS and HERAS spectra, article revision; J.L.R.-S.: experiment conception and design, analysis and interpretation of Raman, SERS and HERAS spectra, article revision and approval of the final version; E.A.-L.: historical research, preparation of the panel painting reproduction and article revision; and A.M.-V.: critical revision of the draft, with the inclusion of relevant new content.

Competing interests

We declare we have no competing interests.

Funding

This research has been supported by CONACYT projects LN 232619, LN 260779 and LN 271614 for LANCIC instrumentation, as well as CONACYT projects CB 239609 and 131944 and PAPIIT UNAM IN402813 that funded this research development. Panel painting reference project was funded by PAPIIT UNAM Project IN402007.