An in-and-out-the-lab Raman spectroscopy study on street art murals from Reggio Emilia in Italy

The street art murals ‘The Big Mother’ by Gola Hundun, the ‘Big Sacral Bird’ by Kenor, the ‘Oriental Carpet” by H101 and “The Economy Subdues You” by Zosen, belonging to the Cooperative Popular Houses of Mancasale and Coviolo in Reggio Emilia (Italy), were investigated by the use of various mobile Raman spectrometers coupled to different lasers and by micro-Raman spectroscopy on selected samples. The study was made necessary by the evident fading of many colours, despite the young age of the paintings, realized in 2010. The first step of the investigation, realized by the on-site campaign, was the identification of the materials, and in particular of the dyes. The main chromophores were identified as polycyclic, monoazo- and disazo- organic pigments, with inorganic compounds as bismuth vanadate (BiVO4) together with the extensive presence of rutile (TiO2). The second step was devoted to the study of the degradation mechanism affecting the colourful layers of the murals. It required the use of laboratory micro-spectrometers and was carried out on a reduced set of samples, selected during the in-situ campaign. This combination of on-site and laboratory Raman spectroscopy allowed the obtaining of the complete identification of the palette used by the different artists in a single day of measurements, in a complete non-destructive day. In addition, it was possible to minimize the number of samples required for the study of the degradation process.


Introduction
Outdoor artworks of contemporary public art are usually realized with modern synthetic materials which suffer from lack of long-last stability, producing alteration effects and deterioration problems according to their chemical composition. Furthermore, the exposure to severe and extreme weather conditions and environmental pollution are crucial issues to be considered in the conservation and protection of these artifacts.
The three-year European project "Conservation of Art in Public Spaces" (CAPuS), involving 15 partners among universities, academies, companies and research centres, is designed to study artworks of contemporary public art, including murals and outdoor sculptures. The CAPuS project, co-funded by the ERASMUS + Knowledge Alliances 2018-2021 programme, aimed at contributing to the dissemination of knowledge in the field of public art conservation. In particular, the CAPuS project has developed specific conservation guidelines for artworks in public spaces together with an innovative training module for higher education institutions as well as an open e-learning module for professionals [10]. An illustrated multi-lingual glossary providing definitions in street art and conservation, and an open-access digital repository containing subpages of selected artworks and specific details (i.e. materials analysis, condition reports, and general archival documentation) have been also realized and constantly implemented during project lifespan and beyond [11]. The characterization of artworks materials and their degradation processes is of fundamental importance to the development of conservation treatment protocols. Within the project, the Italian partners focused on works of post muralism and street art to first study the compositional materials and their characteristic degradation processes and later develop specific conservation methods. The data acquired during in situ campaigns contributed to the formalization of conservation guidelines, as well as to the creation of formative modules for students and professionals. a e-mail: danilo.bersani@unipr.it (corresponding author) 0123456789().: V,-vol

Materials and methods
In situ Raman analysis was performed on the street murals 'The Big Mother" by Gola Hundun, "Big Sacral Bird" by Kenor, 'Oriental Carpet' by H101, and 'The Economy Subdues You' by Zosen. All the mobile Raman spectrometers were brought simultaneously on the field, and the analyses were carried out directly on the colourful surfaces of the street murals without jeopardising the works of art. An aerial work platform was used to measure the big-sized murals 'Big Sacral Bird' and 'The Economy Subdues You' (Fig. 1a and b) (up to 15 m) while a ladder with no additional equipment was used for investigating the 'Oriental Carpet' and 'The Big Mother'. During a single day of on-site analysis, 47 different colour regions have been analysed, covering all the different chromatic hues, apparently stable and unstable, painted both by spray and by roller, of the four wall paintings. Every region was analysed with at least two different spectrometers, in most cases with three spectrometers, repeating the measurements in different (usually 3) spots for each area. A huge amount of spectral data was obtained in a very short time, minimizing the impact on the environment and on the inhabitants of the buildings. The analysed points are reported in Fig. S2 (supplementary material), as well as with the hue changes of the different chromatic areas.
Some millimeter-sized samples were collected during the measurements by means of a scalpel from the positions indicated in Fig. S2 (supplementary material). Four of them were embedded in epoxy resin and polished. The as-obtained samples and the cross-sections were then analyzed by a micro-Raman spectrometer.

In situ and direct analysis
All the mobile Raman spectrometers used for the on-site analysis of the outdoor murals can be visualized in Fig. 1c.

The Bravo handheld Raman spectrometer (Bruker, Ettlingen, Germany)
The Bravo handheld Raman system is a dual laser (785 and 853 nm) compact spectrometer able to acquire spectra in an extended spectral region from 300 to 3200 cm −1 . The system is coupled to a CCD detector, its spectral resolution is 10-12 cm −1 and the laser power is fixed (less than 100 mW). This battery-operated spectrometer that weights 1.5 kg uses an automatic integrated calibration. Bravo is using the sequentially shifted excitation (patented) technology to reconstruct a spectrum that is free from fluorescence effects. The Bravo Raman system is actually collecting slightly shifted Raman spectra from slightly different laser excitations produced by changing the lasers' current. In practice, it uses two spectral regions; from 300 to 2200 cm −1 for the 785 nm laser and from 1200 to 3200 cm −1 for the 853 nm laser, to collect 3 shifted Raman spectra per laser per spectral region and reconstructs a 7th final spectrum free of fluorescence. In theory, if the differences in laser excitations are small (slightly different), the broad fluorescence features can remain unchanged while the actual Raman signal is shifted. The user can separately view all the spectra recorded together with the final one. This is very useful in cases where the in-built algorithm of the instrument produces artefacts other than the actual Raman bands.
All the measurements were conducted in the automatic mode bringing the plastic protective head (used to secure the lens/laser optics) of the fixed probehead of the spectrometer in contact with the surface under study. The working distance is defined at around 4 to 5 mm while the spot size is smaller than 1 mm (both as reported by the manufacturer).

The dispersive i-Raman® EX portable system (BWTek, Newark, USA)
The i-Raman® EX is a fibre optics dispersive Raman spectrometer coupled to a 1064 nm laser and equipped with a TE-cooled InGaAs detector. The system records spectra in the spectral range from 100 to 2500 cm −1 , with a spectral resolution of 10 cm −1 . The adjustable laser power can reach up to 499 mW, while the main unit weighs 3.4 kg. The spectrometer is coupled to an external laptop, to allow for setting the experimental conditions and collecting the spectra. In general, the experimental conditions used were set at 10 accumulations of 6 s with a laser power reaching 50%. In some cases, the total measuring time and laser power were modified (e.g. 20 accumulations of 3 s at 50% or 20% laser power; 10 accumulations of 6 s at 20% laser power) to avoid detector saturation. The objective lens has a 5.9 mm working distance and a light-blocker with a protective foam was slid over the probehead. The spot size is less than 0.5 mm (measured with the light blocker). No post-calibration of the data was performed but the system was checked on the field and prior to the analysis using sulphur (UCB) and cyclohexane (Kaiser).

The RaPort handheld Raman analyzer (EnSpectr, San José CA, USA)
The EnSpectr RaPort Raman spectrometer is equipped with a frequency-doubled Nd:YAG laser with a wavelength of 532 nm. The maximum output power is 30 mW. To avoid thermal damage to the organic colorants, the laser power was lowered via software between 20 and 50% of the maximum. The instrument is handheld, with a weight of 2.5 kg. During the measurement session, the power was supplied by the internal battery, allowing wireless operation up to 6 h. The measurement range is fixed but very wide, from 125 to 4070 cm −1 , with a maximum spectral resolution of 7 cm −1 . The minimum spot size is 0.5 mm. No calibration is required before measurements. Measurement time ranged from 10 to 30 s per spectrum. The instrument is USB-connected to a laptop, where the spectra are visualized and saved in ASCII format for identification and processing.

Micro-Raman spectroscopy
Micro-Raman analysis was performed on all the samples taken during the on-site campaign, without further preparation, and on cross-sections embedded in resin obtained from three samples. Non-polarized Raman spectra were recorded at 632.8 nm (He-Ne laser) in a nearly backscattering geometry with a Horiba LabRam microspectrometer equipped with an integrated Olympus BX40 microscope. The spectral resolution was about 2 cm −1 . The power on the sample was kept under 10 mW using neutral density filters. A 100X objective was used to collect the Raman signal, with a space resolution of 1 µm. A Raman profile containing 20 points with 2 µm of step was obtained on a selected cross-section.

Software
All data acquisition was performed with software provided by the manufacturers. Post manipulation of the in situ Raman spectra was performed with OPUS™ software (Bruker) and Thermo Grams/AI 8.0 ® suite software (Thermo Fischer Scientific). Horiba LabSpec 5 was used for the acquisition and processing of the micro-Raman maps.

In situ Raman spectroscopy: issues and comparison of the instruments
For the purpose of the analysis of the street art belonging to the Cooperative Popular Houses of Mancasale and Coviolo in Reggio Emilia (Italy), three mobile Raman spectrometers, coupled to different laser excitations were brought to the field. Although the direct Raman analysis is the optimal approach for characterizing cultural heritage artefacts without jeopardising them, its application is not always straightforward.
The measurements conducted on the murals colourful surfaces were all concluded in one day during the day time. Raman spectroscopy is challenged by light interferences, and thus, the identification of the unknown can be hampered. In Fig. 2a, it is illustrated that the RaPort Raman system suffered light interferences, when measuring a coloured area (red faded to light red) from the 'Big Sacral Bird'. The excess fluorescence present in the spectrum, made it difficult to identify the synthetic organic pigment present on the painting. The i-Raman ® EX and Bravo Raman spectrometers were able to characterize the main pigment used as the diketopyrolo-pyrole (DPP) pigment PR254 (Fig. 2 b, c and Table 1. When measuring in situ, ambient light interferences can be avoided by ensuring dark conditions between the Raman probehead of the spectrometer and the surface under study. In general, this can be achieved either by measuring at night or in dark conditions, e.g. measuring in light-blocking tents, switching off the lights when measuring in museums, etc. In this case, dark conditions were created by bringing the probeheads in complete contact with the surfaces. For the i-Raman ® EX Raman spectrometers, the long fibre optics probehead can easily reach any distant surface and by sliding a light-blocker cup with a protective foam over the probehead (Fig. 1d) the darkened measuring conditions can be met. Indeed, in none of the spectra collected with this mobile system from the three street art murals, light interferences were observed. In the case of RaPort and Bravo Raman systems, these are coupled to fixed optical heads (Fig. 1e). Although in the Bravo data, no major interferences were observed, in several spectra from RaPort the daylight was reflected in the spectra. Spectra of Bravo seem unaffected to these interferences as these are effectively eliminated by the sequentially shifted excitation algorithm that the spectrometer uses to treat and produce the final spectrum.
Regarding the fluorescence observed, in general, this is unavoidable and it is a phenomenon inherited in Raman spectroscopy. One can use higher laser wavelengths to try to avoid electronic transitions, hence minimise the fluorescence emission. However, by increasing the wavelength, the decreasing of the Raman scattering is expected. RaPort is equipped with a 532 nm laser, a wavelength that often is ideal for resonance Raman spectroscopy [46]. Several components of the coloured surfaces of the murals were identified by using the 532 nm laser, thanks to the resonance of some pigments. Unfortunately, in some cases, fluorescence was overwhelming the Raman signals. In general, as better evidenced in the following, for the analysis of mural paintings the advantage of having a better efficiency by using a short-wavelength (e.g. 532 nm) laser, does not compensate for the big disadvantage of inducing too much fluorescence. The 1064 nm laser of the i-Raman ® EX worked adequately in most of the cases with the materials characterization, although fluorescence still occurred in some cases. Moreover, in order to achieve an acceptable signal-to-noise ratio in a short measurement time, the i-Raman ® EX was operated at 50% of its maximum power. The same situation was valid when using the RaPort spectrometer: a compromise between short acquisition time (essential when using a handheld instrument without tripods or other supports) and a good signal-to-noise ratio, would require the use of elevated laser power. However, excessive power can cause alteration to the measured material, which could be observed as the Raman signal of the colorant was decreasing when increasing laser power, evidencing their photo-degradation [47][48][49]. So the careful choice of the laser power when analysing different colours (different light absorption) was highly important.
To eliminate fluorescence or signals other than the Raman bands, Bravo uses the sequentially shifted excitation technology. For each laser (785 and 853 nm), the instrument acquires three spectra in slightly different excitations in order to reconstruct two fluorescence-free spectra, that are finally merged to a single final spectrum. Indeed, Bravo produced the best results from the mobile Raman systems used in this study and was able to measure all of the high-and most of the low-intensity bands for the components present on the surfaces. Drawbacks of the system are the non-flexible probehead and the fixed total laser power used. Moreover, although this handheld Raman spectrometer can be easily operated by a non-expert, the interpretation of its data is sometimes challenging. One should always cross-check the final spectrum with the raw data, to check for possible spectral artefacts: in some measurements, the algorithm produced an erroneous and difficult to interpret final spectrum.
From all the aforementioned, the necessity of collecting data with several Raman instruments using different lasers and technologies is underlined. Specifically, from the in situ measurements conducted on the outdoor street art murals belonging to the Table 1 Overview of the street art murals' colours together with their materials identification, for all the mobile Raman systems discussed in the study. The colours fading is indicated (when possible) in the colour column. The band positions are referring to the Bravo handheld Raman system and are assigned according to [29-31, 33, 52, 53]. Particular assignments with other Raman systems/lasers are underlined in the band positions column. The band positions in brackets denote unresolved bands. + , positive identification; x, no identification The spectra were easily collected as the measurements were conducted on an automatic mode and the materials identification was more complete compared to i-Raman ® EX and RaPort. The i-Raman ® EX better characterized the components of the murals than the RaPort, with the latter positioned at the bottom place out of the three. Although working on unstable scaffolding with direct sunlight were challenging conditions, the different mobile Raman systems were able to characterize the materials found on the surfaces of: the 'Big Sacral Bird' by Kenor, 'The Big Mother' by Gola Hundun and 'The Economy Subdues You' by Zosen. In Table 1, the identification of materials collected from different areas on the murals, with the three mobile systems used, is described together with the identified Raman band positions for each component.

In situ identification of the chromophores
In all the analysed areas, a suitable dye or pigment was identified, representing the vast majority of the chromophores present in the paintings. The results are summarized in Table 1, with the indication of the instruments able to identify them. In the following, some details of the results obtained on the different paintings are shown.

'Big Sacral Bird' by Kenor
The monoazopigment, acetoacetic arylide PY74 was found on the discoloured surfaces of the 'Big Sacral Bird'. In Fig. 3a, the spectrum of the PY74 of an orange faded to pink area is presented. In this spectrum the bands at 1349 and 1297 cm −1 (indicated in Fig. 3a with an *) are unresolved. Although the low spectral resolution of Bravo might pose several identification problems on bands found in close spectral proximity, as here the bands have a spectral difference of more than 10-12 cm −1 , in this case, the system did not suffer this problem. Thus, other explanations involving a real change in the spectrum should be involved: a small band shift or broadening can be attributed to degradation phenomena that could have occurred naturally (weathering) or due to the laser power. The same unresolved pattern was noticed also in some spectra recorded with the i-Raman ® EX system, although this was not frequently observed, as in most cases these bands were resolved. In the spectrum in Fig. 3b, the bands of PY74 seem more resolved compared to the spectrum of Fig. 3a.
Additional examples of unresolved bands of synthetic organic pigments can be also found in several cases with several Raman systems, with some examples demonstrated in Table 1, and in Figs. 3d, and 4c, d. Again, this might point out to a degradation visualized in the Raman spectra, but the source of the degradation is not straightforward (weathering, photo degradation or postlaser degradation). Ciccola et al. in 2019 [17], when studying the role of SOPs in the degradation of Acrylem AC 33 under UVB, observed certain alterations on the Raman signals before and after exposure. For some SOPs band broadening was observed (while for others not). After averaging spectra collected from Raman mappings and by band fitting, the visualization of changes was possible for the PB 15:1, PG 7 and PR 264 pigments. The Raman bands before and after the degradation experiments were contrasted on the basis of normalization with the unaltered bands. Pause et al. in 2021 [50] evaluated the BRAVO Raman system for the identification of SOPs in varnished paints. Micro-Raman spectroscopy was also used for comparison reasons. The authors, among other aspects regarding Raman spectroscopy in general and the Bravo handheld instrument in particular (fluorescence, spectral resolution, laserinduced damage) are discussing the shift or merge or broadening of the Raman bands, and they attributed that phenomenon to the elevated laser power of the handheld instrument. Although, shifting of the Raman bands may be attributed to other reasons [30,39]. Moreover, broadening and merging can occur when bands are overlapping, especially in cases of molecular vibrations found on close wavenumbers.
For the murals discussed in the current study, if the spectral resolution is not responsible for the unresolved bands, broadening, shifting, and merging can be the final product of either natural exposure to the outdoor environmental conditions or laser degradation. For the Bravo Raman system, the laser power is fixed (less than 100 mW) with literature reporting that laser powers of 45 mW [50] and 50 mW [51] are reaching the sample surface. For the other mobile Raman systems also elevated lasers powers were used. Shifting of the Raman bands is not only so straightforward as it can arise from natural or laser-induced degradation but also from focusing, stability, calibration, sample form, shape or size of the granules or crystals, etc. so no major conclusions are discussed regarding this aspect in the current study.
According to Scherrer et al. [29], PY74 demonstrates strong vibrations at 1593 and at 1352 cm −1 and a very strong band at 1328 cm −1 . In a gold yellow faded to light brown area (Fig. 3c) the combination of the disazopigment, diarylide PY83 and bismuth vanadate (BiVO 4 ) were responsible for the colouring [52,53]. The phthalocyanine pigment PB15 together with bismuth vanadate were responsible for the shade of one of the measured green colours. The symmetric (Ag) and antisymmetric (Bg) bending modes of bismuth vanadate (BiVO 4 ) [53] seemed partially unresolved in the current study. Although, the aforementioned reasons may play a role for the unresolved and broad features, this might also be indicative of the presence of monoclinic and tetragonal BiVO 4 [52]. Also, monoclinic and tetragonal BiVO 4 differ on their symmetric V-O stretching mode (Ag symmetry) with tetragonal sharing a band at 850 cm −1 and monoclinc below 830 cm −1 [53][54][55].
Concerning the red pigments, PR254 (Fig. 2b and c) and the naphthol AS monoazopigment PR112 (Fig. 3d) were positively identified on different areas of the mural, while a disazopigment (pyrazolone PO34) was found for the orange-red faded to brown and Fig. 3 Representative Raman spectra collected from the street mural 'Big Sacral Bird', with the Bravo handheld Raman spectrometer, corresponding to a an orange faded to pink area (PY74 and rutile (R)), b a lemon yellow faded to pale yellow area (PY 74 and rutile (R)) c gold yellow faded to light brown area (PY83, bismuth vanadate (BV) and rutile (R)) and d a red area (PR112) light purple areas. In a deep blue area, the polycyclic dioxazine pigment PV23 was positively identified together with the possible incorporation of PB15.

'The Big Mother' by Gola Hundun
PB15 was found in different blue areas measured on the 'The Big Mother' mural (Fig. 4a). The Bravo Raman system was unable to identify the second pigment present in the mixture. In the data from the blue areas collected from the RaPort spectrometer, a band at ca. 1390 cm −1 is possibly attributed to the presence of PV23. PV23 is the main colour of the dark violet area measured (Fig. 4b).

'The Economy Subdues You' by Zosen
PY74 was positively characterized as an orange faded to light yellow (Fig. 4c) and to white areas, respectively. In Fig. 4d the pigment PO34 was found in a red faded to purple area. Pigment PB15 is responsible for the turquoise hue found on the outdoor mural's surface.

Other components
In most of the areas, measured C-H stretching vibrations were found above 2900 cm −1 , while measuring with the Bravo instrument. Its extended spectral region (until 3200 cm −1 ) and the lasers used are ideal for Raman signals found in this spectral area. A band at ca. 1723 cm −1 present in many spectra (sometimes together with bands at ca. 1042 and 1002 cm −1 ) might be due to the resin used [56]. Alkyd and modified alkyd solvent-based spray paints have been mainly employed by the artists, i.e. MTN 94 spray cans, as suggested by archive documentations [9,43].
Eur. Phys. J. Plus (2022) 137:252 Fig. 4 Representative Raman spectra: collected from the mural 'The Big Mother' with the Bravo handheld Raman system, corresponding to a blue area (PB15), b a dark violet area (PV23, calcite (C) and rutile (R)); collected from the mural 'The Economy Subdues You' with c the RaPort handheld Raman spectrometer, corresponding to an orange faded to light yellow area (PY74 and rutile (R)) and d with the dispersive i-Raman ® EX portable system, corresponding to a red faded to purple area (PO34 and rutile (R)) Calcite (CaCO 3 ) is positively identified in the spectrum of Fig. 4b while barites (BaSO 4 ) can be a possible interpretation of the band at 985 cm −1 (although downshifted and in absence of other bands) from an orange faded to the pink area (Fig. 3a).
Rutile (TiO 2 ) is undoubtedly the protagonist in most of the spectra collected from the 'Big Sacral Bird' and the 'The Big Mother' (only a few did not contain spectral features of rutile) and in all the spectra collected from 'The Economy Subdues You'. As mobile Raman spectroscopy is a surface-sensitive technique, by using in situ analysis it is not possible to determine whether the TiO 2 originates from the paint mixtures or whether it migrated to the surface from an underlying layer. The migration of rutile is a phenomenon that has been demonstrated before on outdoor street murals [9,57]. Possible answers to the presence of rutile were given with the help of micro-Raman spectroscopy performed on cross-sections.

Analysis of degradation and fading
The four investigated artworks, exposed to extreme sunlight and weather conditions, are characterized by visible chromatic alterations and problems of cracking with flaking and losses with the crack pattern that follows the bricks underneath.
'The Big Mother' by Gola Hundun, the 'Big Sacral Bird' by Kenor, the 'Oriental Carpet' by H101 and 'The Economy Subdues You' by Zosen are exposed to sunlight and weather. In 'The Economy Subdues You', the most exposed of the murals (absence of trees and south orientation), the general fading is more evident than in other artworks. In two artworks (by Kenor and H101), there are also visible problems of cracking with flaking and losses with the crack pattern that follows the bricks underneath.
In locations where the prime coating is absent, the damage is slightly more severe, with pronounced cracks, larger flakes, and in the lower part, also subfluorescence between the ground layer and the plaster.
During the condition report compilation and the photographic documentation, recurring alterations on some colors emerged. For example, an orange color used in three artworks has turned into greyish purple. Some orange, purple and yellow colours faded partially, while a dark pink used in almost every artwork disappeared completely.

Micro-Raman spectroscopy
During the measurement campaign performed with mobile instruments, a few millimetre-size painting samples were taken on positions subject to the major colour changes to be studied with micro-Raman, in order to understand the fading mechanism.
Micro-Raman analyses have been carried out on all the samples, without further preparation, showing the same pigments detected with the mobile spectrometers. The rutile underlayer was detected in almost all samples, allowing to better understand the stratigraphy of the painting. From 3 samples (one for each painting, except "The Big Mother") showing the most evident fading issues, crosssections were obtained. In all cases, the point-to-point micro-Raman analysis showed a similar situation: a rutile underlayer, a painting layer containing one or more dyes and minor rutile, and a very thin outer layer with rutile and a small amount of dye. To better understand this phenomenon, a linear Raman micro-map was carried out on the sample "H101" taken from "oriental Carpet", selected as representative for its clear stratigraphy and its big change in hue.

'Oriental Carpet' by H101
The sample "H101" (from the name of the artist) comes from a light pink-purple area of the 'Oriental Carpet' painting. From the photographic record, it is evident that the original hue was an intense orange-red. The cross-section (Fig. 5) reveals the presence of a thin, white ground layer (less than 10 µm) over the plaster, then a thicker vivid orange layer, with variable thickness (average value of ca. 40 µm) and a very thin greyish outer layer (few µm).
Although a limited number of in situ measurements were performed with all the mobile Raman instruments, we are able to confirm the presence of PO34 and rutile in the particular area discussed here. In general, rutile was identified in most of the measured areas.
Micro-Raman measurements, performed in different points of the three layers, demonstrated the presence of rutile in the ground layer and of PO34 in the orange layer, together with a minor amount of rutile. The outer layer, which is clearly related to the faded colour, contains mostly rutile, with minor concentrations of PO34. To better follow the relative concentrations of the two compounds, a micro-Raman profile was obtained along the cross-section, from plaster to the outer layer ( Fig. 6a and b). The Raman profile clearly shows how the amount of rutile (marked with 'R'), maximum in the white preparation layer, increases from the orange layer to the outer grey one. On the contrary, the signal of PO34 (marked with 'P') decreases significantly from the orange layer to the outer one.
The fading is related to the relative enrichment of rutile in the outer layer. Two possible mechanisms could explain this phenomenon. The first one is the photo-degradation of PO34 due to the exposition of the sunlight, depleting the outer layer and producing a local higher concentration of rutile. This is the simplest possible explanation, but this would mean that the lightfastness of PO34 (and many other pigments showing fading in these paintings) is very poor, less than expected. One should take into account the fact that nanocrystalline TiO 2 rutile is an efficient photo-catalyst, able to induce degradation of organic molecules exposed to sunlight [58][59][60][61]. The absence of any PO34 degradation products, which have not been found in the outer layer, can be supported by the efficient photocatalytic activity of TiO 2 . PO34 molecules have been locally degraded by the photo-catalyst into volatile species (H 2 O, CO 2 ) which are no longer on the artwork surface.
The second possible explanation is the migration of rutile from the inner layers to the outer surface. The migration seems to start from the inner layer, because the concentration of rutile in the orange layer is very low. The migration could happen through micro-cracks in the paint layers or through diffusive phenomena that need further investigation. Even if this explanation seems more complex, it appears to better correspond to the observations. The migration of rutile particles through a paint layer was already observed by one of the authors during a master thesis work on the degradation of the Klein Blue, presented at different conferences Further studies on artificially aged synthetic paint layers could help to investigate the mechanism at the basis of the migration of rutile and to better quantify the lightfastness of the involved organic dyes. It should be noted that during the last years, the hypothesis of the migration of rutile particles was already invoked in similar cases [9,57].

Conclusions
In situ Raman spectroscopy and micro-Raman spectroscopy were employed to identify the artists' palette and investigate degradation phenomena affecting these colourful works of art. In particular, the three mobile Raman spectrometers employed, allowed us to identify the vast majority of the pigments used and underline the extensive presence of rutile on the murals' surfaces. Micro-Raman spectroscopy performed on selected cross-sections not only confirmed this but was also able to locate the rutile in the stratigraphy. We notice those recent developments in the use of the Raman spectroscopy, in particular of the SORS (Spatially Offset Raman Spectroscopy) technique, could allow to obtain stratigraphic information on paintings without the need for sampling [34].
'The Big Mother' by Gola Hundun, the 'Big Sacral Bird' by Kenor, the 'Oriental Carpet' by H101 and 'The Economy Subdues You' by Zosen are exposed to sunlight and weather. In 'The Economy Subdues You', the most exposed of the murals (absence of trees and south orientation), the general fading is more evident than in other artworks. In two artworks (by Kenor and H101), there are also visible problems of cracking with flaking and losses with the crack pattern that follows the bricks underneath.
In locations where the prime coating is absent, the damage is slightly more severe, with pronounced cracks, larger flakes, and in the lower part, also subfluorescence between the ground layer and the plaster.
During the condition report compilation and the photographic documentation, recurring alterations on some colors emerged. For example, an orange color used in three artworks has turned into greyish purple. Some orange, purple and yellow colours faded partially, while a dark pink used in almost every artwork disappeared completely.
Taking into account the higher damages, in terms of changes of hue and cracking, experienced by the paintings with higher solar exposition and the results obtained on the stratigraphic section, it is possible to hypothesize that the alterations present are mostly due to the disappearance of the organic pigments present in the spray products (first microns) of the paint layer, leaving the TiO 2 -based extender and therefore returning a gray-violet tone in some cases (due to the underlying orange color still present) or in many other cases a white background. This provides interesting information on spray products and is of great use for their conservation. It can also be used by artists, to avoid unstable tints or use other types of colors such as quartz dispersion or acrylic-siloxane paints. Moreover, the micro-Raman stratigraphic analysis suggests that a second mechanism, the migration of rutile from the ground layer to the surface, can also play an important role in the fading process. In order to understand the relevance of this phenomenon, further investigations are required.
fellowship with project number: 12X1919N. The authors would also like to thank Ylenia Cobelli for her help during the on-site analysis of the outdoor mural paintings and data interpretation, as these were parts of her degree thesis.

Data Availability Statement
This manuscript has associated data in a data repository. [Authors' comment: Data associated with the present work and other investigations related to the CAPuS project can be found at https://www.capusrepository.unito.it/. The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.] Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.