New Advances in Dye Analyses: In Situ Gel-Supported Liquid Extraction from Paint Layers and Textiles for SERS and HPLC-MS/MS Identification

To date, it is still not possible to obtain exhaustive information about organic materials in cultural heritage without sampling. Nonetheless, when studying unique objects with invaluable artistic or historical significance, preserving their integrity is a priority. In particular, organic dye identification is of significant interest for history and conservation research, but it is still hindered by analytes’ low concentration and poor fastness. In this work, a minimally invasive approach for dye identification is presented. The procedure is designed to accompany noninvasive analyses of inorganic substances for comprehensive studies of complex cultural heritage matrices, in compliance with their soundness. Liquid extraction of madder, turmeric, and indigo dyes was performed directly from paint layers and textiles. The extraction was supported by hydrogels, which themselves can undergo multitechnique analyses in the place of samples. After extraction, Ag colloid pastes were applied on the gels for SERS analyses, allowing for the identification of the three dyes. For the HPLC-MS/MS analyses, re-extraction of the dyes was followed by a clean-up step that was successfully applied on madder and turmeric. The colour change perceptivity after extraction was measured with colorimetry. The results showed ΔE values mostly below the upper limit of rigorous colour change, confirming the gentleness of the procedure.


Introduction
While sample reduction and procedure miniaturisation are generally desirable in analytical chemistry, the principle of minimal invasiveness is imperative when analysing cultural heritage and, where possible, completely noninvasive analyses are preferred.
Nonetheless, when sampling is forbidden the study of organic components is disadvantaged. Dyes, in particular, represent one of the most complex challenges because of their low concentration and tendency to fade. When they are in complex matrices, care must be taken not to completely lose information about their presence. Noninvasive techniques, like fibre optic reflectance spectroscopy (FORS) or fluorescence spectroscopy, were proved to be effective, in some instances, for dye identification in rather simple matrices [1][2][3][4][5], but they generally lack specificity. Because of their strong fluorescence, dyes are not suited for Raman spectroscopy identification. However, fluorescence can be quenched when signals are enhanced by means of metal nanoparticles (NPs), and very good dye spectra can be obtained by means of surface enhanced Raman spectroscopy (SERS) [6][7][8][9][10]. Nonetheless, the phenomenon occurs when the analyte interacts with metal NPs, which cannot be applied was also adapted to paint layers and extended to additional dye classes. Thre mock-ups were prepared dying wool using madder, indigo, and a direct dye: t Furthermore, madder lake pigment and indigo in powder were mixed with eg obtain tempera paint mock-ups. While the ammonia-EDTA extraction and dLLM up procedure was used on madder and turmeric, indigo gels were imbibed aqueous solution of NaOH:Na2S2O4 1:2, able to reduce indigotin into its wate form, leuco indigo [16,28] (Figure 1). Once in the hydrogel pores, leuco indigo r into indigotin when in contact with the atmosphere. A dLLME clean-up proced then tested to re-extract indigoids from the gels for the HPLC-MS/MS analyses. A are listed in Section 3 (Materials and Methods).

Madder and Turmeric
The ammonia-EDTA extraction solution was visibly able to extract mad turmeric dyes both from wool and paint layers. As already observed by Germina [19], agar gel was homogeneously coloured after extraction (Figure 2a), whil Nanorestore Gel ® HWR, the dyes appeared more concentrated on the contac (Figure 2b), in accordance with its high retention power.

Madder and Turmeric
The ammonia-EDTA extraction solution was visibly able to extract madder and turmeric dyes both from wool and paint layers. As already observed by Germinario et al. [19], agar gel was homogeneously coloured after extraction (Figure 2a), while for the Nanorestore Gel ® HWR, the dyes appeared more concentrated on the contact surface (Figure 2b), in accordance with its high retention power.
form, leuco indigo [16,28] (Figure 1). Once in the hydrogel pores, leuco ind into indigotin when in contact with the atmosphere. A dLLME clean-up pr then tested to re-extract indigoids from the gels for the HPLC-MS/MS analys are listed in Section 3 (Materials and Methods).

Madder and Turmeric
The ammonia-EDTA extraction solution was visibly able to extract turmeric dyes both from wool and paint layers. As already observed by Ger [19], agar gel was homogeneously coloured after extraction (Figure 2a), w Nanorestore Gel ® HWR, the dyes appeared more concentrated on the co (Figure 2b), in accordance with its high retention power.

Indigo
Agar gel was extremely effective for indigoid dye extraction from both wool and paint mock-ups. The gel appeared greenish coloured right after extraction; however, after contact with the atmosphere, the oxidation of leucoindigo back to indigotin slowly turned the colour back to blue. Conversely, the Nanorestore Gel ® HWR showed not to be compatible with the reducing solution. After soaking, the gel appeared altered by look and by touch. A substance absorbed from the solution or degradation product was observable right at the centre of the gel cylinder (Figure 3), and the Nanorestore Gel ® HWR consistency was hardened.
Agar gel was extremely effective for indigoid dye extraction paint mock-ups. The gel appeared greenish coloured right after ext contact with the atmosphere, the oxidation of leucoindigo back to in the colour back to blue. Conversely, the Nanorestore Gel ® HW compatible with the reducing solution. After soaking, the gel app and by touch. A substance absorbed from the solution or degrada servable right at the centre of the gel cylinder (Figure 3), and the N consistency was hardened.

Madder and Turmeric
Spectra were acquired directly on the gels after the addition o loids were made in pastes to prevent excessive absorption into the tion 3 (Materials and Methods)). The spectra recorded show SER tributable to madder and turmeric dyes (Table 1). Peaks at approxi 1320 cm −1 , observed in the spectra of the agar gel after madder extra paint mock-ups (Figure 4a), are diagnostic for the presence of aliza to the C-C stretching, H-C-C, and C-C-C bending modes of the an 19,29-34].

Madder and Turmeric
Spectra were acquired directly on the gels after the addition of Ag colloids. The colloids were made in pastes to prevent excessive absorption into the gel cylinders (see Section 3 (Materials and Methods)). The spectra recorded show SERS scattering peaks attributable to madder and turmeric dyes (Table 1). Peaks at approximately 1270-1280 and 1320 cm −1 , observed in the spectra of the agar gel after madder extraction from textile and paint mock-ups (Figure 4a), are diagnostic for the presence of alizarin and are attributed to the C-C stretching, H-C-C, and C-C-C bending modes of the anthraquinone ring [15][16][17][18][19][29][30][31][32][33][34]. Table 1. Main SERS scattering peaks observed in the spectra of agar gel (3% w/v) and Nanorestore Gel ® HWR after extraction of madder and turmeric from wool and tempera mock-ups. The vibrational modes are reported as ν (stretching) and δ (bending).

Peaks from Madder
Attribution Peaks from Turmeric Attribution Peaks from the Gel Matrix Tempera  Textile  Tempera  Textile  Textile  Textile  838  loids were made in pastes to prevent excessive absorption into the gel cylinders (see Section 3 (Materials and Methods)). The spectra recorded show SERS scattering peaks attributable to madder and turmeric dyes (Table 1). Peaks at approximately 1270-1280 and 1320 cm −1 , observed in the spectra of the agar gel after madder extraction from textile and paint mock-ups (Figure 4a), are diagnostic for the presence of alizarin and are attributed to the C-C stretching, H-C-C, and C-C-C bending modes of the anthraquinone ring [15][16][17][18][19][29][30][31][32][33][34].
(a) (b)  The peak at approximately 1615 cm −1 , mainly visible in the spectrum from wool extraction, is attributed to the C=O stretching of the anthraquinone ring [14,15,29,30]. The peak at approximately 1150 cm −1 (1144 for tempera and 1155 cm −1 for wool) can be ascribed to the C-C stretching and C-H bending modes of the dye molecules [29]. Additional medium and weak intensity peaks between 800 and 1000 cm −1 , such as the one at 887 cm −1 , are attributed mainly to anthraquinone skeletal vibration [14,15,29]. The peaks at approximately 1450 cm −1 are generally attributed to alizarin C-O stretching and C-O-H and C-H bending [14,15,30]. The scattering band at 1398 cm −1 , on the other hand, is diagnostic for the presence of purpurin [30,34,35]. This peak is reproducibly more pronounced after extraction from wool than from tempera and could be due to the different preparation recipes. Additional peaks, such as the ones at approximately 1360 and 1505 cm −1 , could not be assigned to specific modes of anthraquinone dyes and can, hence, be related to the complex molecular pattern extractable from madder roots. Scattering peaks at approximately 1330-1350 cm −1 , however, were already observed in the agar gel [19,35], and its attribution, hence, remains uncertain.
Madder spectra acquired on Nanorestore Gel ® HWR ( Figure 4b) similarly exhibit peaks at approximately 1155, 1290, 1320, and 1620 cm −1 . The peaks were less intense in comparison to the ones from the gel matrix, especially after extraction from wool. In this latter case, peaks at approximately 1422 and 1449 cm −1 , with a shoulder at 1465 cm −1 , were preferentially enhanced. These peaks are generally attributed to alizarin C-O stretching and C-O-H and C-H bending [14,15,29]. Nonetheless, peaks in the same positions were also observed in the gel matrix. In addition, the Nanorestore Gel ® HWR matrix presents some peaks at approximately 1320-1350 cm −1 that overlap with the ones typical of alizarin. A peak at 1041 cm −1 , visible after extraction from tempera, is attributed in the literature to the alizarin ring C-C-C in-plane bending [29]. The intensity of this specific peak in this spectrum could be due to the geometry of the interaction between the analytes and the AgNPs.
In general, on the paint layers as on the textile, the typical anthraquinone dye SERS scattering peaks were more recognizable after agar-gel-supported extraction. This is in accordance with already published results [19]. This can be related to the high water retention of Nanorestore Gel ® HWR, which could, in some instances, hinder the extraction solution release and, hence, the analytes collection.
In the case of turmeric, too, typical SERS scattering peaks could be observed on the agar gel after extraction from wool ( Figure 5). Signals at 1139 and 1166 cm −1 can be assigned to the molecular skeleton vibration, while the signal at 1290 cm −1 is due to the phenolic ring C-C-C, C-C-H, and C=CH bending [36]. Lastly, the band at 1600 cm −1 can be assigned to the C-OH bending and its shoulder at 1630 cm −1 to the C=C and C=O stretching [36,37]. The peak at 999 cm −1 is likely due to the gel matrix. On the contrary, no peaks related to curcumin could be observed on Nanorestore Gel ® HWR after the dye extraction. As mentioned for madder, the difference in the extraction capabilities can be explained by taking into account the difference in water retention.
Molecules 2023, 28, x FOR PEER REVIEW 6 of 16 As mentioned for madder, the difference in the extraction capabilities can be explained by taking into account the difference in water retention. In general, the gel matrices exhibited rather reproducible signals. While the agar gel peak positions are reproducible, anyways, their intensities can change consistently. This fact is a consequence of the agar gel macromolecular structure, which can have variable interaction with AgNPs.
The signals observed on the Nanorestore Gel ® HWR blank were also quite reproducible during the experiments that were carried out, and the main scattering peaks were at 839 cm −1 , 999 cm −1 , 1355 cm −1 , 1422, and 1448 cm −1 .
To avoid any attribution uncertainty, however, especially in the region around 1320 and 1450 cm −1 , SERS analyses of dyes re-extracted from the gels will be attempted in the future.

Indigo
Because of the degradation that occurred in the reducing solution, the Nanorestore Gel ® HWR was not able to extract indigo. Spectra were, thus, acquired only on the agar gel after indigo extraction and Ag colloid pastes addition.
In the spectra recorded, intense and sharp peaks related to indigoids were visible ( Figure 6 and Table 2). Table 2. Main scattering peaks observed in the spectra of agar gel 3% (w/v) after extraction of indigo from wool and tempera mock-ups. The vibrational modes are reported as ν (stretching) and δ (bending). The intense peaks at 544 and 597 cm −1 are related to the indigotin C=C-CO-C and C-N bending modes and to the C-C and C-N bending modes, respectively. The peak at 1224 cm −1 is due to the molecule N-H and C-H in-plane bending, while the one at 1250 cm −1 is In general, the gel matrices exhibited rather reproducible signals. While the agar gel peak positions are reproducible, anyways, their intensities can change consistently. This fact is a consequence of the agar gel macromolecular structure, which can have variable interaction with AgNPs.

Peaks from Indigo
The signals observed on the Nanorestore Gel ® HWR blank were also quite reproducible during the experiments that were carried out, and the main scattering peaks were at 839 cm −1 , 999 cm −1 , 1355 cm −1 , 1422, and 1448 cm −1 .
To avoid any attribution uncertainty, however, especially in the region around 1320 and 1450 cm −1 , SERS analyses of dyes re-extracted from the gels will be attempted in the future.

Indigo
Because of the degradation that occurred in the reducing solution, the Nanorestore Gel ® HWR was not able to extract indigo. Spectra were, thus, acquired only on the agar gel after indigo extraction and Ag colloid pastes addition.
In the spectra recorded, intense and sharp peaks related to indigoids were visible ( Figure 6 and Table 2). The SERS peaks obtained are consistent with the ones published by Platania et al. [16], who extracted indigo from cotton using agar gel loaded with the same reducing solution (NaOH:Na2S2O4 1:2). Nonetheless, in comparison to the spectrum obtained by Platania et al., the spectra reported here exhibit a pronounced enhancement of the peaks at 545, 597, and 1573 cm −1 . Interestingly, the spectra obtained after extraction from wool Figure 6. Spectra of agar gel (3% w/v) cylinders after 5 min extraction of indigo from tempera and from wool. Table 2. Main scattering peaks observed in the spectra of agar gel 3% (w/v) after extraction of indigo from wool and tempera mock-ups. The vibrational modes are reported as ν (stretching) and δ (bending). The intense peaks at 544 and 597 cm −1 are related to the indigotin C=C-CO-C and C-N bending modes and to the C-C and C-N bending modes, respectively. The peak at 1224 cm −1 is due to the molecule N-H and C-H in-plane bending, while the one at 1250 cm −1 is due to the C-H, C=C, and N-H in-plane bending [16]. The strong signal at 1574 cm −1 can be attributed to the molecule's C=C and C=O stretching modes [16,38].

Peaks from Indigo
The SERS peaks obtained are consistent with the ones published by Platania et al. [16], who extracted indigo from cotton using agar gel loaded with the same reducing solution (NaOH:Na 2 S 2 O 4 1:2). Nonetheless, in comparison to the spectrum obtained by Platania et al., the spectra reported here exhibit a pronounced enhancement of the peaks at 545, 597, and 1573 cm −1 . Interestingly, the spectra obtained after extraction from wool and after extraction from tempera are very similar in spite of the different preparation recipes and are highly reproducible.

Madder and Turmeric
The chromatograms obtained show the presence of the target analytes selected. The chromatographic peaks related to alizarin fragmentation (Rt = 4.17 min) and purpurin fragmentation (Rt = 4.52) could be observed after re-extraction from agar gel applied on wool (Figures 7 and 8). The SERS peaks obtained are consistent with the ones published by Platania et al. [16], who extracted indigo from cotton using agar gel loaded with the same reducing solution (NaOH:Na2S2O4 1:2). Nonetheless, in comparison to the spectrum obtained by Platania et al., the spectra reported here exhibit a pronounced enhancement of the peaks at 545, 597, and 1573 cm −1 . Interestingly, the spectra obtained after extraction from wool and after extraction from tempera are very similar in spite of the different preparation recipes and are highly reproducible.

Madder and Turmeric
The chromatograms obtained show the presence of the target analytes selected. The chromatographic peaks related to alizarin fragmentation (Rt = 4.17 min) and purpurin fragmentation (Rt = 4.52) could be observed after re-extraction from agar gel applied on wool (Figures 7 and 8). On the contrary, re-extraction from Nanorestore Gel ® did not produce satisfying results. The dLLME clean-up procedure, successfully tested and validated for aglycones and glycosylated dyes, demonstrated efficacy also with turmeric. The chromatographic peaks related to curcumin transitions (Rt = 4.68) could be observed after re-extraction from agar gel and Nanorestore Gel ® HWR. No peaks related to the analyte were observed in the reference blanks.

Indigo
Because of the extraction solution complexity, the clean-up procedure developed for indigoids (see Section 3 (Materials and Methods)) could not be perfected. The formation of a sulphur salt was observed both in the aqueous phase and organic phase after dLLME For this reason, HPLC-MS/MS analyses of indigoid dyes with this procedure requires further studies, and the protocol is still under refinement. Considering the good results ob tained by means of SERS, however, the reducing solution could be further diluted to hinder the salt precipitation.

Madder and Turmeric
The colour variations induced using gel-supported liquid extraction are mainly around the colour tolerance applied in the industrial field (ΔE = 3 CIELAB units) consid ered the upper limit of rigorous colour tolerance (Table 3) [39].
On madder, only Nanorestore Gel ® HWR extraction from tempera paint induced a colour variation higher than three. However, no mark was visible on the paint to the authors' eyes. Agar gel, on the other hand, induced a colour variation of 2.97 CIELAB units and showed better results when analysed by means of SERS. On wool, both Nanorestore Gel ® HWR and agar gel extraction was invisible, leaving a large margin for further devel opments. On the contrary, re-extraction from Nanorestore Gel ® did not produce satisfying results. The chromatographic peak areas of the analytes are greater for agar gel (3% w/v) extraction than for Nanorestore Gel ® HWR extraction. Again, this is in accordance with their different water retention.
Regarding the extraction from tempera paint layers, no peaks related to alizarin and purpurin were observed. Nonetheless, considering the SERS results obtained, the extraction times from the paint mock-ups could be adjusted to further increase the amount of analyte collected.
The dLLME clean-up procedure, successfully tested and validated for aglycones and glycosylated dyes, demonstrated efficacy also with turmeric. The chromatographic peaks related to curcumin transitions (Rt = 4.68) could be observed after re-extraction from agar gel and Nanorestore Gel ® HWR. No peaks related to the analyte were observed in the reference blanks.

Indigo
Because of the extraction solution complexity, the clean-up procedure developed for indigoids (see Section 3 (Materials and Methods)) could not be perfected. The formation of a sulphur salt was observed both in the aqueous phase and organic phase after dLLME. For this reason, HPLC-MS/MS analyses of indigoid dyes with this procedure requires further studies, and the protocol is still under refinement. Considering the good results obtained by means of SERS, however, the reducing solution could be further diluted to hinder the salt precipitation.

Madder and Turmeric
The colour variations induced using gel-supported liquid extraction are mainly around the colour tolerance applied in the industrial field (∆E = 3 CIELAB units) considered the upper limit of rigorous colour tolerance (Table 3) [39]. The reducing solution proved to be harsher than the ammonia-EDTA solution. The extraction time was reduced from 3 h to 5 min on wool, providing very good SERS results and a colour variation of only 1.52 CIELAB units (Table 3). Conversely, the extraction on tempera induced a colour variation of 4.14 CIELAB after 5 min. Considering the spectra obtained, the reducing solution could be further diluted and the extraction times adjusted.

R1R
Tempera paint, madder lake pigment The reducing solution proved to be harsher than the ammonia-EDTA solution. The extraction time was reduced from 3 h to 5 min on wool, providing very good SERS results and a colour variation of only 1.52 CIELAB units (Table 3). Conversely, the extraction on tempera induced a colour variation of 4.14 CIELAB after 5 min. Considering the spectra obtained, the reducing solution could be further diluted and the extraction times adjusted.

CUR
Wool dyed with turmeric Nanorestore Gel ® HWR 3 h 5.24 extraction time was reduced from 3 h to 5 min on wool, providing very good SERS results and a colour variation of only 1.52 CIELAB units (Table 3). Conversely, the extraction on tempera induced a colour variation of 4.14 CIELAB after 5 min. Considering the spectra obtained, the reducing solution could be further diluted and the extraction times adjusted.

I1AR
Tempera paint, indigo over azurite Agar gel (3% w/v) 5 min 4.14 extraction time was reduced from 3 h to 5 min on wool, providing very good SERS results and a colour variation of only 1.52 CIELAB units (Table 3). Conversely, the extraction on tempera induced a colour variation of 4.14 CIELAB after 5 min. Considering the spectra obtained, the reducing solution could be further diluted and the extraction times adjusted.

IND Wool dyed with indigo
Agar gel (3% w/v) 5 min 1.52 On madder, only Nanorestore Gel ® HWR extraction from tempera paint induced a colour variation higher than three. However, no mark was visible on the paint to the authors' eyes. Agar gel, on the other hand, induced a colour variation of 2.97 CIELAB units, and showed better results when analysed by means of SERS. On wool, both Nanorestore Gel ® HWR and agar gel extraction was invisible, leaving a large margin for further developments.
Turmeric colour measurements showed that the dye is sensitive to the extraction solution. After 3 h, a light shift to a reddish colour was effectively visible under careful observation. Nevertheless, the presence of turmeric could be detected using both SERS and HPLC-MS/MS, which means there could be a margin for a reduction in the extraction time.

Indigo
The reducing solution proved to be harsher than the ammonia-EDTA solution. The extraction time was reduced from 3 h to 5 min on wool, providing very good SERS results and a colour variation of only 1.52 CIELAB units (Table 3). Conversely, the extraction on tempera induced a colour variation of 4.14 CIELAB after 5 min. Considering the spectra obtained, the reducing solution could be further diluted and the extraction times adjusted.

Lake Pigments Preparation
Red madder lake pigment was prepared following a recipe from Daniels et al. [40]. Briefly, 5 g of madder roots were soaked in 150 mL distilled water and left overnight. The roots in water were heated up until 70 • C for 30 min. After filtration, 2.5 g potassium alum was added to the solution, and the temperature was brought to 80 • C. Meanwhile, 0.94 g K 2 CO 3 was dissolved in 25 mL water and gently poured into the dye bath under continuous stirring. The lake pigment formed was left to precipitate overnight. Once precipitated, the lake pigment was filtered and finely ground.

Paint Mock-Ups Preparation
Paint mock-ups were made on bricks prepared with eight layers of CaSO 4 and rabbit skin glue and applied with a brush [41]. The organic pigments were mixed with a solution of 2 mL water and an egg yolk. The pigments were mixed with the binder in a rough proportion 2:1 (w/v) using a spatula and then applied by means of a brush, and then left to age naturally for six months.

Wool Dyeing
Textile mock-ups were prepared from wool dyed using madder, indigo, and turmeric. First, 2 g wool yarn was mordanted as follows: 620 mg alum and 120 mg potassium bitartrate (KC 4 H 5 O 6 ) were dissolved in 250 mL of distilled water. The solution was warmed to 40 • C and kept for ten minutes. Wool was soaked in the solution once cooled down. The temperature was increased again to 80 • C for 40 min and kept for 1 h under gentle magnetic stirring [27]. Once at ambient T, the wool was removed, squeezed, and left to dry.
To dye wool using madder, 1 g madder roots was soaked in 400 mL distilled water and warmed to 35 • C. One gram of mordanted wool was soaked, the temperature was brought to 80 • C over 40 min, and then kept for 1 h under gentle magnetic stirring [27].
For direct wool dyeing using turmeric, 0.5 g turmeric in powder was soaked in 400 mL distilled water, brought to 90 • C, and then kept for 15 min under magnetic stirring. Successively, 1 g unmordanted wool was soaked and left for 30 min at the same temperature.
To dye wool with indigo, 0.6 g ground indigo was added to 10 mL distilled water at 45 • C. Successively, a solution of 0.6 g NaCO 3 in 6 mL distilled water and a solution of 1.5 g Na 2 S 2 O 4 in 50 mL distilled water at 45 • C were added one after the other. The mixture obtained was brought to 55 • C and kept for 20 min. Afterwards, 3 g unmordanted wool was soaked in the bath and left for 10 min. The wool was then removed and left in contact with the atmosphere to allow the dye to re-oxidise.
After the dyeing was performed, distilled water was always used to wash the wool until the rinse was transparent; then, it was left to dry and age naturally for six months. The textile mock-ups were made by winding dyed wool yarns around microscope glass slides.

Agar Gel Preparation
For the agar gel preparation, 0.24 g agar in powder was added to 8 mL distilled water in a 100 mL beaker. The beaker was then gently shaken in a boiling water bath for 10 min so that the polymer reached its melting temperature. The solution was cooled for 30 min at ambient temperature and stored in a refrigerator overnight [19,35].

Ag Colloids Pastes Preparation
The Ag colloids were prepared following the protocol developed by Leopold and Lendl [19,42]. Specifically, 20 mg NaOH was added to 5 mL MilliQ water, and 21 mg NH 2 OH · HCl was added to another 5 mL MilliQ water. The two solutions were mixed together and poured into a solution of 17 mg AgNO 3 in 90 mL MilliQ water under constant magnetic stirring. The obtained Ag colloid solution was stored in the fridge. To obtain colloidal pastes, 10 mL colloids were centrifuged at 4500 RPM for 20 min, and the supernatant was discarded [43,44].

Gel-Supported Liquid Extraction
Agar gel 3% (w/v) and Nanorestore Gel ® HWR, as a commercial product, were cut into cylinders of approximately 4 mm in diameter. The cylinders were soaked for 90 min in a solution of NH 3 (30-33%) and Na 2 EDTA 1 mM (1:1). NaCl was added until a final concentration of 4.7 mM [19,20,27]. After 90 min, the gel cylinders were removed using tweezers and left to lose 5% of their weight. In the case of indigo extraction, after soaking in the reducing solution, the gel cylinders were quickly dried on absorbent paper and applied directly on the mock-ups to avoid prolongated contact with O 2 in the atmosphere. The gel cylinders were applied on paint and textiles with glass slides on the top to prevent solution's evaporation. The extraction time varied, from 5 min on the paint layers to 3 h on the textiles. The whole procedure workflow is summarised in Figure 9 as a schematic diagram.
magnetic stirring. The obtained Ag colloid solution was stored in the fridge. To o colloidal pastes, 10 mL colloids were centrifuged at 4500 RPM for 20 min, and the su natant was discarded [43,44].

Gel-Supported Liquid Extraction
Agar gel 3% (w/v) and Nanorestore Gel ® HWR, as a commercial product, wer into cylinders of approximately 4 mm in diameter. The cylinders were soaked for 90 in a solution of NH3 (30-33%) and Na2EDTA 1 mM (1:1). NaCl was added until a concentration of 4.7 mM [19,20,27]. After 90 min, the gel cylinders were removed u tweezers and left to lose 5% of their weight. In the case of indigo extraction, after soa in the reducing solution, the gel cylinders were quickly dried on absorbent paper applied directly on the mock-ups to avoid prolongated contact with O2 in the atmosp The gel cylinders were applied on paint and textiles with glass slides on the top to pre solution's evaporation. The extraction time varied, from 5 min on the paint layers t on the textiles. The whole procedure workflow is summarised in Figure 9 as a schem diagram.

SERS Analysis
For the SERS analyses, 20 µL of colloidal pastes were poured on the gel face, w was in contact with the mock-ups. The gels were then left to dry for 12 h at ambient perature. The SERS analyses were carried out directly on the dry gel using a Horiba J Yvon HR Evolution micro-Raman spectrometer equipped with a He-Ne laser (λ = 633 coupled with a microscope with a set of interchangeable objectives. The spectra wer lected using a 100× magnification objective, and the laser intensity was varied from to 0.75 mW. The acquisition time was varied from 5 to 10 s and the scan number fro to 60, depending on the sample, in order to obtain the best signal-to-noise ratio. A m mum of three spectra were collected for every sample, both on the reference gel so into the extraction solution (blank gel matrix) and the gel after the dyes' extraction spectra were processed using OriginPro 9 software (©OriginLab): the background subtracted fitting a polynomial baseline to the power of five, the spectra were norma and "adjacent averaging" smoothing was applied to reduce noise.

SERS Analysis
For the SERS analyses, 20 µL of colloidal pastes were poured on the gel face, which was in contact with the mock-ups. The gels were then left to dry for 12 h at ambient temperature. The SERS analyses were carried out directly on the dry gel using a Horiba Jobin-Yvon HR Evolution micro-Raman spectrometer equipped with a He-Ne laser (λ = 633 nm) coupled with a microscope with a set of interchangeable objectives. The spectra were collected using a 100× magnification objective, and the laser intensity was varied from 0.15 to 0.75 mW. The acquisition time was varied from 5 to 10 s and the scan number from 30 to 60, depending on the sample, in order to obtain the best signal-to-noise ratio. A minimum of three spectra were collected for every sample, both on the reference gel soaked into the extraction solution (blank gel matrix) and the gel after the dyes' extraction. All spectra were processed using OriginPro 9 software (©OriginLab): the background was subtracted fitting a polynomial baseline to the power of five, the spectra were normalised, and "adjacent averaging" smoothing was applied to reduce noise.

dLLME
The dLLME clean-up procedure was developed and validated to enhance dyes recovery. The development and validation study is under submission to another journal by the same authors. After the gel-supported liquid extraction from the mock-ups, the dyes were re-extracted from the gels for the HPLC-MS/MS analyses using the same aqueous solution. The gels loaded with madder, and the turmeric dyes were soaked in 0.8 mL NH 3 (30-33%), 0.8 mL Na 2 EDTA (1 mM), and 4.4 mg NaCl. After 24 h, 495.6 mg NaCl was added together with 1 mL HCl 6 M and 0.8 mL HCOOH (≥95%) to bring the solution pH to 3. The dyes were then extracted from the aqueous phase into the organic solvent: 250 µL 2-propanol was added to every sample and, subsequently, together in the same syringe, 200 µL 1-pentanol and 100 µL 2-propanol were vigorously injected to obtain a highly dispersed thee-phasic system, known as a cloudy solution [28,45,46].
For indigo dyes, the gels were re-extracted in 1.5 mL distilled water containing NaOH:Na 2 S 2 O 4 1:2 (w/w). After 24 h, 530 µL HCl 6 M was added, and the solution volume was brought to 5 mL with distilled water. For the extraction into organic solvent, 750 µL 2-propanol and 100 µL chloroform were vigorously injected together to obtain the cloudy solution [46]. All samples were successively sonicated for 10 min and centrifuged for 10 min at 10,000 RPM and 5 • C. The aqueous solution was discarded. For samples containing anthraquinone dyes, 1-pentanol was washed using 3.1 mL of a solution of 2.8 M NaCl. Lastly, the organic solvent was evaporated under N 2 flow. All samples were reconstituted with 100 µL MeOH:H 2 O 1:1 for the HPLC-MS/MS analyses.

HPLC-MS/MS Analyses
For the chromatographic analysis, a SCIEX Exion LC AD System was used. The system was coupled to a Sciex QTRAP 6500 mass spectrometer system with electrospray ionisation (ESI) and multicomponent IonDrive Technology.
The column chosen was a reversed phase BEH C18 (2.1 mm × 50 mm) with 1.7 µm silica particles. The injected volume was 3 µL. The mobile phases chosen were 0.1% formic acid in Millipore water (phase A) and 0.1% formic acid in acetonitrile (phase B). The gradient programme is shown in Table 4. Table 5 reports the MRM transitions used for the main target analytes' identification (in negative) based on optimisation using certified standards. Identifications were made relying on retention times and two fragmentation transitions.

Colorimetric Measurements
To assess the perceptivity of the procedure on paint layers and textiles, colorimetric measurements were performed before and after gel-supported extraction, and the ∆E values were calculated. Colorimetric coordinates were acquired using fibre optic reflectance spectroscopy (FORS) in the visible range. The spectrophotometer used was a BWTEK Exemplar LS (B&W Tek, Plainsboro, NJ, USA) with a tungsten lamp BWTEK (series BPS101, 5 W, emission spectrum from 350 to 2600 nm and 2800 K colour temperature). The acquisition range was from 180 to 1100 nm, with a resolution from 0.6 nm to 6.0 nm. The fibre optic, a THORLABS RP22, was used with a head to obtain a 45 • inclination of the probe. The measurements were performed in the dark, before and after gel extraction, on the same point. Every measurement was repeated three times and mediated. The CIE L*a*b* parameters were extracted from the spectra using the software BWSpec. The colour variation ∆E00 was calculated using the formula reported in the CIEDE2000 guidelines [39,50].

Conclusions
In conclusion, gel-supported in situ extraction of madder, turmeric, and indigo was applied for the first time on wool and tempera paints for SERS and HPLC-MS/MS dyes identification.
The ammonia-EDTA solution proved to be able to extract turmeric, a direct dye, from textiles. Furthermore, the solution, already tested on water soluble paint layers [36], allowed anthraquinone dye extraction also from tempera paints that were naturally aged for six months.
The dyes extracted using the ammonia-EDTA solution were identified directly on the gels after Ag colloid pastes addition by means of SERS. While madder-related peaks were clearly distinguishable on both agar gel and Nanorestore Gel ® HWR, no turmeric-related peaks could be detected on Nanorestore Gel ® HWR.
A reducing solution, already tested by Platania et al. for gel-supported extraction from cotton [16], was selected for indigoids. Nanorestore Gel ® HWR demonstrated to not be compatible with the reducing solution, while agar gel extraction was successful on wool and tempera paints. The SERS analysis on agar gel produced very good results, and indigoids-related peaks were clearly recognizable and highly reproducible.
A dLLME clean-up procedure was applied for the HPLC-MS/MS analyses. The procedure was necessary for the dye re-extraction from the gels and purification before injection in the instrument.
The dLLME clean-up workflow was already applied by Serafini et al. for the HPLC-MS/MS identification of anthraquinone dyes extracted from ancient textiles using the ammonia-EDTA solution. The results are under submission to another journal. The procedure demonstrated, for the first time, to be effective also for the HPLC-MS/MS identification of turmeric.
Moreover, a clean-up protocol for HPLC-MS/MS analysis of indigoids is still under development. Further dilution of the reducing solution could be effective in perfecting the dLLME procedure tested, which could not be applied because of a salt precipitation.
For both spectroscopic and chromatographic analyses, dye-related signals observed after Nanorestore Gel ® HWR extraction were less intense when compared to the ones observed after agar gel (3% w/v) extraction. This is in accordance with the Nanorestore Gel ® HWR retention power. Hence, unless dealing with very water sensitive materials [36], agar gel use is preferable.
The colour change results on the mock-ups before and after gel-supported liquid extraction were very promising for all dyes tested. Considering the SERS and HPLC-MS/MS results, there is still a margin for improvement. The gel cylinder dimension could be consistently reduced and extraction times extended to make the procedure completely imperceptible.
Future research will be focused on artificially aged mock-ups to assess the effect of degradation phenomena on invasiveness and extraction efficiency.
Ultimately, the presented gel-supported in situ extraction is suited for multianalytical dye identification. Techniques traditionally used to study dyes, both invasively and noninvasively, can be combined without interferences from the matrix. During HPLC-MS/MS analyses, the addition of further transitions to the instrumental parameters related to the analytes of interest can allow for a more accurate portrayal of the dyes' molecular pattern, including glycosylated moieties [28]. Thus, the approach constitutes a valid alternative to accompany noninvasive analyses of inorganic substances for comprehensive studies of cultural heritage without posing a threat to their integrity.