Historical Pigments and Paint Layers: Raman Spectral Library with 852 nm Excitation Laser

: Raman spectroscopy (RS), for its robust analytical capabilities under constant development, is a powerful method for the identification of various materials, in particular pigments in cultural heritage. Characterization of the artist’s palette is of fundamental importance for the correct formulation of restoration intervention as well as for preventive conservation of artworks. Here we examine the number and variability of research studies exploiting Bravo handheld Raman spectrophotometer relying on the excitation of Raman signal with temperature-shifted diode lasers emitting at 852 and 785 nm. To this end, we explore the spectral features of common historical pigments examined as powders and in the paint layer. We show that some materials may exhibit slightly different spectra as concerns especially the relative intensity of Raman lines with 852 nm laser excitation wavelength as compared to the standard 785 nm. The aim is to provide the research community with a reference spectral database that facilitates the identification of unknown pigments using the 852 nm excitation source.


Introduction
The identification of pigments in artworks may provide relevant information to date/detect forgeries, establish the artwork's origin, or formulate suitable restoration and preventive conservation procedures.Raman spectroscopy is nowadays a well-established method in the field of heritage science for its capacity to identify non-destructively and non-invasively a large variety of art materials [1].The development of increasingly efficient low-size and lightweight Raman micro-probes enables in situ measurements and investigation of the artwork without any movement, sampling, or damage [2].A robust spectral database of reference substances might ease the interpretation and exploitation of Raman spectral information, recently also through automated processes.Numerous Raman spectral databases have been developed (Table 1) with the first online databases of Raman spectra libraries dating back to the 1990s.Among the most frequently exploited libraries, we can list those provided by University College London (UCL) [3], Infrared and Raman Users Group (IRUG) [4], INFRA-ART by the Romanian database of Raman spectroscopy [5], RRUFF™ by University of Arizona [6], Infrared and Raman Discussion Group (IRDG) [7], SOPRANO [8], and Cultural Heritage Science Open Source (CHSOS) [9].The last two databases are particularly focused on pigments of historical and artistic interest.In addition to the digitally available spectral datasets, publications containing Raman spectra compendia on art-related materials are available in the literature [10].Despite the Minerals 2024, 14, 557 3 of 27 Photoluminescence from either the sample or the substrate may produce signals on the order of or sometimes greater than the Raman signal itself.These phenomena can have an impact on the final Raman spectra, depending on the employed laser wavelength, highlighting or hiding chemical information from the sample.As a consequence, the Raman spectrum of one pigment can appear different depending on whether the laser excitation is in resonance with an electronic transition or whether other photoluminescence bands appear.The literature reporting on the spectral pigments databases includes laser excitation sources, such as Ar+ (488 and 514 nm) or Kr+ (531 and 647 nm), He:Ne (633 nm), Nd:YAG (532 and 1064 nm), and diode laser (630 and 785 nm) [27].
The recently released handheld spectrophotometer by Bruker Optics, commercially named Bravo, employs two different excitation lasers (852 and 785 nm) to cover a spectral range initially comprised between 300 and 3200 cm −1 [28].After 2020, the instrument spectral range was extended to cover a wider range comprised between 170 and 3200 cm −1 , which was a significant improvement, particularly for the investigation of inorganic compounds.The sequentially shifted excitation technology (SSE TM ) and the PCA-based algorithm are exploited to provide the processed background-free and smoothed spectra.This processing procedure is particularly helpful in solving the common background collection produced by the short optical paths typical of portable instrumentation [29], as well as from the strong luminescence produced by cultural heritage organic materials.The application of Bravo Raman has increased considerably in the heritage science field as is evidenced by the recently published scientific literature (Figure 1).The trend of the employment of the instrumentation is exponential, with a decrease in 2020 corresponding to the COVID-19 pandemic.Bravo spectrophotometer proved powerful in addressing research problems pertinent to the identification of historical and modern pigments and lakes , minerals and gems [22,29,54], manuscripts [55,56], mosaics [2,57], and metal and metal corrosion products [58][59][60].The present work aims to create a database using the 852 nm laser excitation wavelength for the most commonly employed art pigments, which can be regarded as the precursor of a larger open-source database of artistic colorants that is expected to be available online in the upcoming years.The goals of this study are to verify the possible differences in the Raman spectra acquired at 852 nm as compared to 785 nm when analyzing a set of different historical pigments frequently encountered during the analysis of polychrome surfaces, and based on the spectral characteristics of the materials, generate a summary to support researchers and conservators in the interpretation of the Raman spectra.

Samples
A total of 32 pure pigments have been analyzed as dry powder.The analyzed pigments are divided into color categories.Table 2 lists all pigments with additional information.The samples represent a selection of the major pigments utilized in artworks.Additionally, with the aim to verify the applicability of the spectral references of pure pigments using the 852 nm laser excitation wavelength of the Bravo instrument, a mock-up panel painting, prepared in 1995 at the Opificio delle Pietre Dure (Florence, Italy) following the historical recipe for egg-tempera, was analyzed.Also, a comparison with spectra obtained with a 785 nm laser excitation wavelength allowed us to identify changes in the position and relative intensity of the characteristic peaks of some pigments in the spectra obtained with the 852 nm laser excitation wavelength.

Samples
A total of 32 pure pigments have been analyzed as dry powder.The analyzed pigments are divided into color categories.Table 2 lists all pigments with additional information.The samples represent a selection of the major pigments utilized in artworks.

Samples
A total of 32 pure pigments have been analyzed as dry powder.The analyzed pigments are divided into color categories.Table 2 lists all pigments with additional information.The samples represent a selection of the major pigments utilized in artworks.

Samples
A total of 32 pure pigments have been analyzed as dry powder.Th pigments are divided into color categories.

Samples
A total of 32 pure pigments have been analyzed as dry powder.T pigments are divided into color categories.

Samples
A total of 32 pure pigments have been analyzed as dry powder.T pigments are divided into color categories.

Samples
A total of 32 pure pigments have been analyzed as dry powder.T pigments are divided into color categories.

Samples
A total of 32 pure pigments have been analyzed as dry powder.Th pigments are divided into color categories.

Mock-Up Panel Painting
Restorers at Opificio delle Pietre Dure (Florence, Italy) realized the mo painting in 1995.It was prepared to employ wooden support with a ground la glue and gypsum (CaSO4•2H2O) (ratio 1:16 v/v) until saturation of the glue, a a brush in two steps (the second layer was applied once the first one was com On top of the ground layer, the imprimitura made of rabbit glue and H2O in a was applied.Once completed, the prepared surface was divided into sections slots (1 cm × 3 cm).Egg tempera was prepared by mixing two parts of yolk white and one of vinegar.The mock-up panel painting (Figure 2) includes blu

Mock-Up Panel Painting
Restorers at Opificio delle Pietre Dure (Florence, Italy) realized the mo painting in 1995.It was prepared to employ wooden support with a ground la glue and gypsum (CaSO4•2H2O) (ratio 1:16 v/v) until saturation of the glue, a a brush in two steps (the second layer was applied once the first one was com On top of the ground layer, the imprimitura made of rabbit glue and H2O in a was applied.Once completed, the prepared surface was divided into sections slots (1 cm × 3 cm).Egg tempera was prepared by mixing two parts of yolk white and one of vinegar.The mock-up panel painting (Figure 2

Mock-Up Panel Painting
Restorers at Opificio delle Pietre Dure (Florence, Italy) realized the moc painting in 1995.It was prepared to employ wooden support with a ground lay glue and gypsum (CaSO4•2H2O) (ratio 1:16 v/v) until saturation of the glue, a a brush in two steps (the second layer was applied once the first one was comp On top of the ground layer, the imprimitura made of rabbit glue and H2O in a was applied.Once completed, the prepared surface was divided into sections slots (1 cm × 3 cm).Egg tempera was prepared by mixing two parts of yolk, white and one of vinegar.The mock-up panel painting (Figure 2) includes blu

Mock-Up Panel Painting
Restorers at Opificio delle Pietre Dure (Florence, Italy) realized the mo painting in 1995.It was prepared to employ wooden support with a ground la glue and gypsum (CaSO4•2H2O) (ratio 1:16 v/v) until saturation of the glue, a a brush in two steps (the second layer was applied once the first one was com On top of the ground layer, the imprimitura made of rabbit glue and H2O in a was applied.Once completed, the prepared surface was divided into sections slots (1 cm × 3 cm).Egg tempera was prepared by mixing two parts of yolk white and one of vinegar.The mock-up panel painting (Figure 2) includes bl

Mock-Up Panel Painting
Restorers at Opificio delle Pietre Dure (Florence, Italy) realized the mock-up panel painting in 1995.It was prepared to employ wooden support with a ground layer of rabbit glue and gypsum (CaSO 4 •2H 2 O) (ratio 1:16 v/v) until saturation of the glue, applied with a brush in two steps (the second layer was applied once the first one was completely dry).On top of the ground layer, the imprimitura made of rabbit glue and H 2 O in a ratio of 1:32 was applied.Once completed, the prepared surface was divided into sections by drawing slots (1 cm × 3 cm).Egg tempera was prepared by mixing two parts of yolk, one of egg white and one of vinegar.The mock-up panel painting (Figure 2) includes blue pigments such as azurite, indigo, Lapis lazuli, smalt, cobalt blue, Prussian blue, and artificial ultramarine and was investigated using conditions specified in Table 3.

Mock-Up Panel Painting
Restorers at Opificio delle Pietre Dure (Florence, Italy) realized the mock-up panel painting in 1995.It was prepared to employ wooden support with a ground layer of rabbit glue and gypsum (CaSO4•2H2O) (ratio 1:16 v/v) until saturation of the glue, applied with a brush in two steps (the second layer was applied once the first one was completely dry).On top of the ground layer, the imprimitura made of rabbit glue and H2O in a ratio of 1:32 was applied.Once completed, the prepared surface was divided into sections by drawing slots (1 cm × 3 cm).Egg tempera was prepared by mixing two parts of yolk, one of egg white and one of vinegar.The mock-up panel painting (Figure 2) includes blue pigments such as azurite, indigo, Lapis lazuli, smalt, cobalt blue, Prussian blue, and artificial ultramarine and was investigated using conditions specified in Table 3.

Sequentially Shifted Excitation (SSE) Raman Spectroscopy
The Bravo spectrometer uses a patented technology called SSE™ (Sequentially Shifted Excitation, patent number US8570507B1) to mitigate fluorescence.The device is equipped with two excitation lasers (DuoLaserTM, 785 nm, and 852 nm) that can be temperatureshifted three times over a small wavelength range (about 0.4 nm).Both laser beams impinge on the sample sequentially in every measurement, covering a broad spectral range from 170-2000 cm −1 and 2000-3200 cm −1 , exploiting the 852 nm and 785 nm lasers, respectively.The laser power was set automatically, reaching a maximum of 100 mW for an 852 nm laser, in an area spot of about 100 µm × 500 µm [45].In this work, we considered the spectral range 170-2200 cm −1 , corresponding to the 852 nm laser excitation, and only the first channel raw data (called CH2) are reported.
The pigment powders were placed on a sheet of aluminum foil, and the Bravo instrument was placed in a vertical position over two aluminum square supports (Figure 3).In this configuration, the laser beam impinges vertically on the pigment powders without any interference.
Minerals 2024, 13, x FOR PEER REVIEW 8 of 27 beams impinge on the sample sequentially in every measurement, covering a broad spectral range from 170-2000 cm −1 and 2000-3200 cm −1 , exploiting the 852 nm and 785 nm lasers, respectively.The laser power was set automatically, reaching a maximum of 100 mW for an 852 nm laser, in an area spot of about 100 µm × 500 µm [45].In this work, we considered the spectral range 170-2200 cm −1 , corresponding to the 852 nm laser excitation, and only the first channel raw data (called CH2) are reported.
The pigment powders were placed on a sheet of aluminum foil, and the Bravo instrument was placed in a vertical position over two aluminum square supports (Figure 3).In this configuration, the laser beam impinges vertically on the pigment powders without any interference.The measurements were first performed with parameters automatically set by the instrument.These are determined by the internal algorithm that evaluates the signal-tonoise ratio.Starting from the so obtained values of integration time, the number of scans was further increased to improve the quality of the obtained spectrum (i.e., 0.5-50 s detector integration time and 5-30 accumulations-Table 3) depending on the response of the investigated materials.

Micro-Raman Renishaw (785 nm)
A benchtop Raman confocal microscope (Renishaw inVia) equipped with a Leica DM2700 optical microscope was employed to acquire micro-Raman spectra using a 785 nm excitation diode laser.We performed the measurement in the spectral range 170-1800 cm −1 , using a grating 1200 lines/mm and a thermoelectrically cooled CCD pixel (functional resolution 400-1060 cm −1 ).The laser powder was kept below 7 mW, using a 10 s exposure time and 5 accumulations.

Stereomicroscopy
The powder pigments were examined under a Leica M205C stereomicroscope with a camera Leica DFC 295 at 0.78×.White balance over a white surface was performed before capturing the images that were processed with LAS v4.6 software.

Results
Raman spectra acquired with an 852 nm laser excitation wavelength of each pigment are reported here.The results are conventionally organized into sections according to the The measurements were first performed with parameters automatically set by the instrument.These are determined by the internal algorithm that evaluates the signalto-noise ratio.Starting from the so obtained values of integration time, the number of scans was further increased to improve the quality of the obtained spectrum (i.e., 0.5-50 s detector integration time and 5-30 accumulations-Table 3) depending on the response of the investigated materials.

Micro-Raman Renishaw (785 nm)
A benchtop Raman confocal microscope (Renishaw inVia) equipped with a Leica DM2700 optical microscope was employed to acquire micro-Raman spectra using a 785 nm excitation diode laser.We performed the measurement in the spectral range 170-1800 cm −1 , using a grating 1200 lines/mm and a thermoelectrically cooled CCD pixel (functional resolution 400-1060 cm −1 ).The laser powder was kept below 7 mW, using a 10 s exposure time and 5 accumulations.

Stereomicroscopy
The powder pigments were examined under a Leica M205C stereomicroscope with a camera Leica DFC 295 at 0.78×.White balance over a white surface was performed before capturing the images that were processed with LAS v4.6 software.

Results
Raman spectra acquired with an 852 nm laser excitation wavelength of each pigment are reported here.The results are conventionally organized into sections according to the pigments' color as follows: 3.1 blue, 3.2 green, 3.3 yellow, 3.4 red, and 3.5 white and black pigments.A summary of the peaks' assignment with relevant references is reported in Table A1 of Appendix A. The Raman spectra obtained also with 785 nm laser excitation wavelength are reported for comparison (in green lines) only when notable differences in Raman peak position/intensity or spectral shape have been encountered.

Blue Pigments
The 852 nm Raman spectra of eight blue pigments-cobalt blue, Prussian blue, Antwerp blue, ultramarine, azurite, indigo, smalt, and Lapis lazuli-are displayed respectively in Figure 4a-h.For Cobalt blue, smalt, and Lapis lazuli, some differences between 852 and 785 nm-generated Raman spectra were detected; therefore, comparative spectra are reported in Figure 4a,e,f.
Cobalt blue (CoO•Al 2 O 3 , Figure 4a) is detected by the bands positioned at 204 cm −1 (Co-O tetrahedral sites vibrational), 410 cm −1 (symmetric bending of AlO 4 ), 514 cm −1 (asymmetric stretching of AlO 4 ), 615 cm −1 (antisymmetric stretching of AlO 4 ), and 754 cm −1 (symmetric stretching vibration of the AlO 4 ) within the spinel lattice [64].As compared to the spectrum with the 785 nm laser excitation line (Figure 4a, green line), the number of bands and their positions are the same; however, their intensities at the low wavenumbers are inverted, being the ratio 204/410 cm −1 lower for the 852 nm laser.Lapis lazuli, as ultramarine, is detected by the band at 548 cm −1 and a shoulder at 583 cm −1 (Figure 4h) attributed to lazurite S3 − and S 2 − symmetric stretching mode.On the contrary to the spectrum of the artificial pigment (Figure 4c) and to that obtained with 785 nm (Figure 4h), this spectrum obtained through 852 nm laser excitation wavelength is very complex and characterized by several sharp and broad bands (namely at 312 vs, 350 m, 405 w, 513 m, 712 w, 835 s, 985 m), not correlated directly to lazurite.This spectral behavior is probably due to luminescence phenomena.In general, the activators of luminescence phenomena-including rare earth elements, Fe 2+ , Fe 3+ , Co 2+ , and Ni 2+ -are chemical impurities in the mineral structure.The 785 nm laser excitation wavelength Raman spectrum gave rise to two sharp bands, one related to lazurite (548 cm −1 ) and the luminescent band at 1306 cm −1 attributed to the presence of diopside [73].Gonzáles-Cabrera et al. [74] exploit these luminescence phenomena to discriminate natural from synthetic ultramarine blue.Other researchers [75,76] link these wavelength-dependent luminescence spectral patterns to the geographical provenance of the samples.It is most likely that the observed luminescent pattern at 852 nm laser excitation wavelength is associated with diopside; however, more in-depth research is necessary to explore its precise origin.To showcase a different pigment spectral behavior under different laser excitation wavelengths, the example of Lapis lazuli is reported in Table 4. Lapis lazuli is detected by the band at 548 cm −1 referred to mineral of lazurite in the spectra acquired with all the laser excitation wavelengths.The spectra generated with 532 nm and 638 nm laser excitation wavelengths, in addition to 260 a cm −1 and 548 cm −1 (bending and stretching of S3 − ion, respectively), also show overtone bands (548 × 2 = 1096 cm −1 , 548 × 3 = 1644 cm −1 , etc.) or combinations [260 + 548 = 808 cm −1 ; 260 + (2 × 548) = 1358 cm −1 ; etc.].The shoulders at ca 258 and 583 cm −1 correspond to asymmetric modes of lazurite.This Raman fingerprint is typical of resonance Raman spectra of lazurite [77].
Other laser excitation wavelengths, i.e., 785 nm, 852 nm, and 1064 nm, do not generate resonance Raman spectra and exhibit a lower number of bands.The Raman spectrum acquired with 1064 nm laser excitation wavelength is very noisy, and only 364 cm −1 and 548 cm −1 peaks are detected.The Raman spectrum of Prussian blue (Figure 4b) is detected by the two low-frequency bands located at 275 cm −1 and 538 cm −1 due to Fe-CN-Fe deformation vibration and Fe-C stretching vibration, respectively [65].The major bands of Prussian blue, positioned at 2090 and 2159 cm −1 (C≡N stretching vibration), are absent in the 852 nm generated spectrum owing to the detection range limits; however, they can be detected by the other 785 nm laser excitation wavelength of instrumentation (not reported in Figure 4b).
Antwerp blue (Figure 4c) is a combination of Prussian blue and Cobalt blue [66].The Raman spectra obtained with 852 nm laser excitation wavelength highlight only the presence of Prussian blue for the two low-frequency bands located at 275 and 533 cm −1 , and no information about the presence of cobalt blue was detected [65].
Ultramarine (Figure 4d), a synthetic pigment of an approximate formula Na 6-10 Al 6 Si 6 O 24 S 2-4 , is identified by its principal band at 548 cm −1 (and its shoulder at 585 cm −1 ) due to the symmetric stretching mode of S 3− in the sulfur-containing sodium-silicate pigment.Its first overtone at 1096 cm −1 is also detected [67].
The overall intensity and quality of the Raman spectrum of azurite at 852 nm is low (Figure 4e).In agreement with the spectra obtained at 785 nm, it is characterized by a band placed at 404 cm −1 (lattice mode), by the bands related to the symmetric and asymmetric stretching of carbonate at 1095 cm −1 and 1425 cm −1 , respectively, and by the symmetric bending mode detectable at 835 cm −1 .Low-intensity bands present at lower frequencies are assigned to lattice modes (220 and 284 cm −1 ).Furthermore, the out-of-plane bending mode of the OH group present in the azurite molecule is located at 939 cm −1 [68].In the considered spectral range, however, it is not possible to detect the bands at 3427 and 3453 cm −1 assigned to the hydroxyl-stretching modes of the OH unit of azurite [69].
The Raman spectrum of smalt (Figure 4g) is dominated by the intense band assigned to the Si-O-Si vibration mode of silicate at 689 cm −1 [71], and the lower frequencies are assigned to the breathing vibration mode of the T-O-T substructure and correlate with the size of the TO 4 tetrahedra ring, where T = Si or Al [72].Similarly to Cobalt blue, as compared to the spectrum with a 785 nm laser excitation line (Figure 4g, green line), the band intensities at the low wavenumbers are inverted, being the 195/481 cm −1 ratio lower for the 852 nm laser.
Lapis lazuli, as ultramarine, is detected by the band at 548 cm −1 and a shoulder at 583 cm −1 (Figure 4h) attributed to lazurite S 3 − and S2 − symmetric stretching mode.On the contrary to the spectrum of the artificial pigment (Figure 4c) and to that obtained with 785 nm (Figure 4h), this spectrum obtained through 852 nm laser excitation wavelength is very complex and characterized by several sharp and broad bands (namely at 312 vs, 350 m, 405 w, 513 m, 712 w, 835 s, 985 m), not correlated directly to lazurite.This spectral behavior is probably due to luminescence phenomena.In general, the activators of luminescence phenomena-including rare earth elements, Fe 2+ , Fe 3+ , Co 2+ , and Ni 2+ -are chemical impurities in the mineral structure.The 785 nm laser excitation wavelength Raman spectrum gave rise to two sharp bands, one related to lazurite (548 cm −1 ) and the luminescent band at 1306 cm −1 attributed to the presence of diopside [73].Gonzáles-Cabrera et al. [74] exploit these luminescence phenomena to discriminate natural from synthetic ultramarine blue.Other researchers [75,76] link these wavelength-dependent luminescence spectral patterns to the geographical provenance of the samples.It is most likely that the observed luminescent pattern at 852 nm laser excitation wavelength is associated with diopside; however, more in-depth research is necessary to explore its precise origin.
To showcase a different pigment spectral behavior under different laser excitation wavelengths, the example of Lapis lazuli is reported in Table 4. Lapis lazuli is detected by the band at 548 cm −1 referred to mineral of lazurite in the spectra acquired with all the laser excitation wavelengths.The spectra generated with 532 nm and 638 nm laser excitation wavelengths, in addition to 260 a cm −1 and 548 cm −1 (bending and stretching of S 3 − ion, respectively), also show overtone bands (548 × 2 = 1096 cm −1 , 548 × 3 = 1644 cm −1 , etc.) or combinations [260 + 548 = 808 cm −1 ; 260 + (2 × 548) = 1358 cm −1 ; etc.].The shoulders at ca 258 and 583 cm −1 correspond to asymmetric modes of lazurite.This Raman fingerprint is typical of resonance Raman spectra of lazurite [77].Other laser excitation wavelengths, i.e., 785 nm, 852 nm, and 1064 nm, do not generate resonance Raman spectra and exhibit a lower number of bands.The Raman spectrum acquired with 1064 nm laser excitation wavelength is very noisy, and only 364 cm −1 and 548 cm −1 peaks are detected.

Green Pigments
Five green pigments-green earth, cobalt green, chromium oxide, malachite, and viridian-were analyzed.The 852 nm laser excitation spectra are reported in Figure 5a-e.Spectra with 785 nm are reported for green earth (Figure 5a) and viridian (Figure 5d).
Raman spectrum of green earth (Figure 5a) is characterized by several broad bands that cover the spectral range of 200-550 cm −1 .The Raman bands centered at 200 and 277 cm −1 are due to internal vibrations of the MoO 6 octahedra, where Mo is the interlayer metal atom, and those located at 393 and 533 cm −1 are the bands mainly due to the vibrational modes of the SiO 4 tetrahedra [20].These bands indicate the presence of celadonite, as the main mineral component of green earth.In these spectra, because of the natural origin of the pigment, Raman bands of gypsum are also present (1006 and 1135 cm −1 ) [78].By changing the excitation to 785 nm (Figure 5a, green line), an apparent shift towards higher wavenumbers of the 533 cm −1 band is observed (the shift is from 533 to 547 cm −1 ), probably due to intensity changes in an unresolved doublet.Similar behavior was previously reported by Ospitali et al. when changing the excitation from 514.5 to 780.0 nm [79].
Viridian (Figure 5d) is a mixture of anhydrous and hydrated chromium oxides, as the recorded Raman bands at 487 and 584 cm −1 are attributed to the hydrated oxide and that at 262 cm −1 to the anhydrous one [81].On the other hand, the spectrum obtained with 785 nm laser excitation wavelength yields a large fluorescence background and accompanying noise; therefore, its identification is not possible.
Malachite (Figure 5e) is detected by the band at 430 cm −1 , together with the peaks at 219, 270, and 350 cm −1 that are related to lattice modes and bands due to symmetric stretching (1095 cm −1 ) and symmetric bending mode (816 and 834 cm −1 ) of CO3 2− [69].In this spectrum, unlike the one from azurite, the band at 1425 cm −1 (asymmetric stretching vibration) is not detected, and hence, the spectral range reported is cut off at 1200 cm −1 .Malachite (Figure 5e) is detected by the band at 430 cm −1 , together with the peaks at 219, 270, and 350 cm −1 that are related to lattice modes and bands due to symmetric stretching (1095 cm −1 ) and symmetric bending mode (816 and 834 cm −1 ) of CO 3 2− [69].In this spectrum, unlike the one from azurite, the band at 1425 cm −1 (asymmetric stretching vibration) is not detected, and hence, the spectral range reported is cut off at 1200 cm −1 .

Yellow Pigments
The 852 nm Raman spectra of six yellow pigments investigated-massicot, lead-tin yellow type I, Naples yellow, chromium oxide, cadmium yellow, yellow ochre, and raw sienna-are reported in Figure 6a-g.No significant spectral changes are recorded compared with the 785 nm laser excitation wavelength.
In general, the low wavenumber region provides a confident recognition of massicot (PbO); in particular, the major band at 142 cm −1 (symmetric stretching of Pb-O) [45] is not detected by the Bravo instrument due to limited spectral region (>170 cm −1 ).However, the diagnostic bands of massicot at 290 and 385 cm −1 are well discerned (Figure 6a).
Lead-tin yellow type I (Pb 2 SnO 4 ) spectrum (Figure 6b) at 852 nm is well characterized by the bands at 197, 275, 290, 455, and 524 cm −1 .The main band of lead-tin yellow at 130 cm −1 (due to symmetric stretching vibration of Pb-O) [82] is out of the detectable spectral range of the instrument.No significant differences as compared to 785 nm were observed.
Naples yellow (Pb 2 Sb 2 O 7 ) (Figure 6c) has a typical spectral pattern in the 200-400 cm −1 spectral region (290, 318, and 348 cm −1 ) of the vibrational mode of both Sb-O and Pb-O bonds; instead, the band at 510 cm −1 is due to Sb-O stretching of the SbO 6 octahedra [83].Other characteristic bands of Naples yellow are placed at 440 and 613 cm −1 .
Chrome yellow (PbCrO 4 , crocoite, Figure 6d) is detected by the Raman instrument at 852 nm, unveiling the most informative 774-942 cm −1 spectral range related to symmetric stretching of the Cr-O bond [45] detected at 841 cm −1 , together with the bands around 360 cm −1 .
Cadmium yellow (CdS, ZnO) (Figure 6e) is characterized by the bands at 298 and 597 cm −1 assigned to the longitudinal optic phonon (LO) + 2E2 and the overtone 2LO + 2E2 of the CdS crystal lattice [84].The band at 212 cm −1 is related to the longitudinal acoustic (LA) phonon modes of the ZnS crystal lattice [85].The unlabeled bands at 987 cm −1 and 457 cm −1 correspond to the symmetric stretching mode of barite SO 4 tetrahedra, probably added as a filler.

Yellow Pigments
The 852 nm Raman spectra of six yellow pigments investigated-massicot, lead-tin yellow type I, Naples yellow, chromium oxide, cadmium yellow, yellow ochre, and raw sienna-are reported in Figure 6a-g.No significant spectral changes are recorded compared with the 785 nm laser excitation wavelength.
In general, the low wavenumber region provides a confident recognition of massicot (PbO); in particular, the major band at 142 cm −1 (symmetric stretching of Pb-O) [45] is not detected by the Bravo instrument due to limited spectral region (>170 cm −1 ).However, the diagnostic bands of massicot at 290 and 385 cm −1 are well discerned (Figure 6a).
Lead-tin yellow type I (Pb2SnO4) spectrum (Figure 6b) at 852 nm is well characterized by the bands at 197, 275, 290, 455, and 524 cm −1 .The main band of lead-tin yellow at 130 cm −1 (due to symmetric stretching vibration of Pb-O) [82] is out of the detectable spectral range of the instrument.No significant differences as compared to 785 nm were observed.
Naples yellow (Pb2Sb2O7) (Figure 6c) has a typical spectral pattern in the 200-400 cm −1 spectral region (290, 318, and 348 cm −1 ) of the vibrational mode of both Sb-O and Pb-O bonds; instead, the band at 510 cm −1 is due to Sb-O stretching of the SbO6 octahedra [83].Other characteristic bands of Naples yellow are placed at 440 and 613 cm −1 .
Chrome yellow (PbCrO4, crocoite, Figure 6d) is detected by the Raman instrument at 852 nm, unveiling the most informative 774-942 cm −1 spectral range related to symmetric stretching of the Cr-O bond [45] detected at 841 cm −1 , together with the bands around 360 cm −1 .
Cadmium yellow (CdS, ZnO) (Figure 6e) is characterized by the bands at 298 and 597 cm −1 assigned to the longitudinal optic phonon (LO) + 2E2 and the overtone 2LO + 2E2 of the CdS crystal lattice [84].The band at 212 cm −1 is related to the longitudinal acoustic (LA) phonon modes of the ZnS crystal lattice [85].The unlabeled bands at 987 cm −1 and 457 cm −1 correspond to the symmetric stretching mode of barite SO4 tetrahedra, probably added as a filler.

Red Pigments
The 852 nm Raman spectra of six red pigments-caput mortuum, umber, red bole, cinnabar, Carmin naccarat, and Alizarin crimson-are reported in Figure 7a-f.As for the yellow pigments, also the Raman spectra of red pigments at 852 nm are very similar to those generated with a 785 nm laser.
are assigned to HgS stretching modes and are orientation-dependent [87].
Raman spectrum of Alizarin crimson (Figure 7f) is characterized by bands placed at 1480, 1328, and 1292 cm

White and Black Pigments
The spectra of the five white pigments investigated-lead white, gypsum, calcite, barite, and zinc oxide-are reported in Figure 7a-d together with that of ivory black (Figure 7f).Lead white (PbCO3)2•Pb(OH)2) (Figure 8a) is detected by the Raman band at 1051 cm −1 (symmetric stretching of CO3 2− ion) [89] and by the bands at 1365 and 1484 cm −1 due 2− (e) (f)

Mock-Up Panel Painting
To test the applicability of the 852 nm laser excitation wavelength of the Bravo instrument in possible real cases, a wooden panel tempera painting with blue pigments (Figure 2) was analyzed.Raman spectra (raw CH2) acquired directly on the mock-up panel

Mock-Up Panel Painting
To test the applicability of the 852 nm laser excitation wavelength of the Bravo instrument in possible real cases, a wooden panel tempera painting with blue pigments (Figure 2) was analyzed.Raman spectra (raw CH 2 ) acquired directly on the mock-up panel painting are compared with reference powder pigments in Figure 9, showing a good spectral correspondence.located in proximity to the characteristic peaks arising from gypsum in the ground layer.Despite this inconvenience, it is possible to identify the peaks of cobalt blue.
Ultramarine pigment (Figure 10b) is well detected in the mock-up by the band at 548 cm −1 ; interestingly, the spectrum is free from the photoluminescence phenomena as observed earlier for pigment powder.For the Prussian blue layer, the entire spectrum (obtained by merging the spectra from 852 nm and 785 nm excitation laser) is reported in Figure 10c.In fact, the major bands of Prussian blue are placed at 2095 and 2150 cm −1 , as observed for both the mock-up and the pigment powder.The azurite paint layer (Figure 9a) is characterized by the main bands at 404 and 320 cm −1 , the spectrum obtained from the mock-up being more intense.Indigo (Figure 9b) in the mock-up is detected by its major band at 1571 cm −1 , and the spectrum is dominated by the bands of gypsum (marked with red dots in the spectrum) from the ground layer.
The spectrum of Lapis lazuli (Figure 9c) is characterized by the band of lazurite at 548 cm −1 .Other photoluminescence signals are observed as discussed earlier (Figure 4h).The results obtained from the smalt paint layer (Figure 9d) contain all the characteristic bands of the pigments in addition to those arising from the preparation layer.
Among all the spectra registered from the paint model, the spectrum from cobalt blue (Figure 10a) is the most challenging to interpret because the bands from the pigment are located in proximity to the characteristic peaks arising from gypsum in the ground layer.Despite this inconvenience, it is possible to identify the peaks of cobalt blue.

Conclusions
The creation and the availability of a spectral database of reference materials are crucial to enable conservation scientists to rapidly identify and study unknown pigments commonly employed in artworks.This work provides the Raman spectra of 32 pigments of historic and artistic importance acquired at 852 nm laser excitation wavelength.This wavelength proved suitable for the detection of all the pigments investigated even though some of them present a strong fluorescence background.Indeed, despite the weak signal in the Raman spectra of azurite, malachite, viridian, and cadmium yellow, their identification possibly outperforms the 785 nm laser excitation wavelength in some cases.
Some differences in the spectral shape and relative intensity of the Raman peaks between spectra obtained with 852 nm and 785 nm laser excitation wavelength were detected for some blue and green pigments.The results suggest that the spectra of pigments Ultramarine pigment (Figure 10b) is well detected in the mock-up by the band at 548 cm −1 ; interestingly, the spectrum is free from the photoluminescence phenomena as observed earlier for pigment powder.For the Prussian blue layer, the entire spectrum (obtained by merging the spectra from 852 nm and 785 nm excitation laser) is reported in Figure 10c.In fact, the major bands of Prussian blue are placed at 2095 and 2150 cm −1 , as observed for both the mock-up and the pigment powder.

Conclusions
The creation and the availability of a spectral database of reference materials are crucial to enable conservation scientists to rapidly identify and study unknown pigments commonly employed in artworks.This work provides the Raman spectra of 32 pigments of historic and artistic importance acquired at 852 nm laser excitation wavelength.This wavelength proved suitable for the detection of all the pigments investigated even though some of them present a strong fluorescence background.Indeed, despite the weak signal in the Raman spectra of azurite, malachite, viridian, and cadmium yellow, their identification possibly outperforms the 785 nm laser excitation wavelength in some cases.Some differences in the spectral shape and relative intensity of the Raman peaks between spectra obtained with 852 nm and 785 nm laser excitation wavelength were detected for some blue and green pigments.The results suggest that the spectra of pigments such as cobalt blue and smalt exhibit inversions in the relative intensity of the bands at low wavenumbers.An interesting case is represented by Lapis lazuli that upon 852 nm laser wavelength excitation produces a characteristic photoluminescence pattern that differs from that of synthetic ultramarine.The comparison between the two laser excitation wavelengths also highlighted the main advantage of using longer wavelengths for the detection of viridian.The 785 nm spectra of viridian are completely masked by fluorescence; while with 852 nm laser excitation wavelength, the main bands are detectable.
Among the aspects to be considered is the covered spectral range by the 852 nm excitation laser (170-2000 cm −1 ) that excludes the main bands of some pigments, e.g., Prussian blue, massicot, and lead-tin yellow.Anyway, the presence of the other characteristic bands of each pigment leads to their identification.Also considering the 785 nm excitation laser employed by Bravo, the spectral range is extended to 3200 cm −1 enabling to identify the Prussian blue uniquely.
Our future prospects concern the expansion of this online database with other natural and synthetic pigments as well as organic dyes.

Minerals 2024 , 27 Figure 1 .
Figure 1.Number of papers exploiting the Bravo spectrometer in heritage science from its marketing year to date.

Figure 1 .
Figure 1.Number of papers exploiting the Bravo spectrometer in heritage science from its marketing year to date.

Table 2 .Figure 1 .
Figure 1.Number of papers exploiting the Bravo spectrometer in heritage science from its marketing year to date.

Table 2 .Figure 1 .
Figure 1.Number of papers exploiting the Bravo spectrometer in heritage science from year to date.

Figure 1 .
Figure 1.Number of papers exploiting the Bravo spectrometer in heritage science from year to date.

Figure 1 .
Figure 1.Number of papers exploiting the Bravo spectrometer in heritage science from year to date.

Figure 1 .
Figure 1.Number of papers exploiting the Bravo spectrometer in heritage science from year to date.

Figure 1 .
Figure 1.Number of papers exploiting the Bravo spectrometer in heritage science from year to date.

Figure 3 .
Figure 3.The overall set-up for Raman measurements of pigments placed on a sheet of aluminum (left) and shown in detail (right).

Figure 3 .
Figure 3.The overall set-up for Raman measurements of pigments placed on a sheet of aluminum (left) and shown in detail (right).

Figure 5 .
Figure 5. Raman spectra of green pigments: (a) green earth, (b) cobalt green, (c) chromium oxide, (d) viridian, and (e) malachite.Raman spectra acquired with 852 nm laser excitation wavelength are reported in black lines, while those with 785 nm are in green lines (a,d).

Figure 5 .
Figure 5. Raman spectra of green pigments: (a) green earth, (b) cobalt green, (c) chromium oxide, (d) viridian, and (e) malachite.Raman spectra acquired with 852 nm laser excitation wavelength are reported in black lines, while those with 785 nm are in green lines (a,d).

Figure 8 .
Figure 8. Raman spectra of white and black pigments: (a) lead white, (b) gypsum, (c) calcite, (d) barite, (e) zinc oxide, and (f) ivory black.The asterisk indicates a band due to the glass of the tip.

Figure 8 .
Figure 8. Raman spectra of white and black pigments: (a) lead white, (b) gypsum, (c) calcite, (d) barite, (e) zinc oxide, and (f) ivory black.The asterisk indicates a band due to the glass of the tip.

Figure 9 .
Figure 9. Raman spectra of blue pigments in mock-up panel painting: (a) azurite, (b) indigo, (c) lapis lazuli, and (d) smalt.Pink lines (named mock-up) indicate the Raman spectra acquired on mock-up panel painting, and cyan lines (named powder) refer to Raman spectra of pigments in powder.All the spectra are acquired with an 852 nm laser excitation wavelength.

Figure 9 .Figure 10 .
Figure 9. Raman spectra of blue pigments in mock-up panel painting: (a) azurite, (b) indigo, (c) lapis lazuli, and (d) smalt.Pink lines (named mock-up) indicate the Raman spectra acquired on mock-up panel painting, and cyan lines (named powder) refer to Raman spectra of pigments in powder.All the spectra are acquired with an 852 nm laser excitation wavelength.

Figure 10 .
Figure 10.Raman spectra of blue pigments in mock-up panel painting: (a) cobalt blue, (b) ultramarine blue, and (c) Prussian blue (spectral range 170-2500 cm −1 ).Pink lines (named mock-up) indicate the Raman spectra acquired on mock-up panel painting, and cyan lines (named powder) refer to the Raman spectra of pigments in powder.All the spectra are acquired with an 832 nm laser excitation wavelength, and Prussian blue is also acquired with a 785 nm laser excitation wavelength.
Table 2 lists all pigments with information.The samples represent a selection of the major pigments utilized

Table 2 .
Analyzed pigments: name and stereoscopic image, chemical composition, co number, supplier, and article number.The scale bar shown in the first row is valid for images.
Table 2 lists all pigments wi information.The samples represent a selection of the major pigments utilized

Table 2 .
Analyzed pigments: name and stereoscopic image, chemical composition, c number, supplier, and article number.The scale bar shown in the first row is valid fo images.
Table 2 lists all pigments wi information.The samples represent a selection of the major pigments utilized

Table 2 .
Analyzed pigments: name and stereoscopic image, chemical composition, co number, supplier, and article number.The scale bar shown in the first row is valid fo images.

Table 2
lists all pigments wit information.The samples represent a selection of the major pigments utilized

Table 2 .
Analyzed pigments: name and stereoscopic image, chemical composition, co number, supplier, and article number.The scale bar shown in the first row is valid fo images.

Table 2
lists all pigments with information.The samples represent a selection of the major pigments utilized

Table 2 .
Analyzed pigments: name and stereoscopic image, chemical composition, col number, supplier, and article number.The scale bar shown in the first row is valid for images.

Table 3 .
Experimental conditions (spectral range, detector integration time, and number of scans) for the samples in powder and mock-up panel painting for the measurement with the pSSE instrument.

Table 4 .
Peak frequencies and their relative intensities of Lapis lazuli under different laser excitation wavelengths.

and Relative Intensity at 852 nm of Laser Excitation Raman Assignments Figure
s Fe-O-Fe/-OH 549 ν as Fe-OH